r263048 - ARM & AArch64: convert asm tests to LLVM IR and restrict optimizations.
Tim Northover via cfe-commits
cfe-commits at lists.llvm.org
Wed Mar 9 10:54:43 PST 2016
Modified: cfe/trunk/test/CodeGen/aarch64-neon-scalar-x-indexed-elem.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/aarch64-neon-scalar-x-indexed-elem.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/aarch64-neon-scalar-x-indexed-elem.c (original)
+++ cfe/trunk/test/CodeGen/aarch64-neon-scalar-x-indexed-elem.c Wed Mar 9 12:54:42 2016
@@ -1,259 +1,509 @@
-// REQUIRES: aarch64-registered-target
// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon -target-cpu cyclone \
-// RUN: -ffp-contract=fast -S -O3 -o - %s | FileCheck %s
+// RUN: -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s
// Test new aarch64 intrinsics and types
#include <arm_neon.h>
+// CHECK-LABEL: define float @test_vmuls_lane_f32(float %a, <2 x float> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <2 x float> [[TMP1]], i32 1
+// CHECK: [[MUL:%.*]] = fmul float %a, [[VGET_LANE]]
+// CHECK: ret float [[MUL]]
float32_t test_vmuls_lane_f32(float32_t a, float32x2_t b) {
- // CHECK-LABEL: test_vmuls_lane_f32
return vmuls_lane_f32(a, b, 1);
- // CHECK: fmul {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[1]
}
+// CHECK-LABEL: define double @test_vmuld_lane_f64(double %a, <1 x double> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x double> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x double>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <1 x double> [[TMP1]], i32 0
+// CHECK: [[MUL:%.*]] = fmul double %a, [[VGET_LANE]]
+// CHECK: ret double [[MUL]]
float64_t test_vmuld_lane_f64(float64_t a, float64x1_t b) {
- // CHECK-LABEL: test_vmuld_lane_f64
return vmuld_lane_f64(a, b, 0);
- // CHECK: fmul {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+.d\[0\]|d[0-9]+}}
}
+// CHECK-LABEL: define float @test_vmuls_laneq_f32(float %a, <4 x float> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <4 x float> [[TMP1]], i32 3
+// CHECK: [[MUL:%.*]] = fmul float %a, [[VGETQ_LANE]]
+// CHECK: ret float [[MUL]]
float32_t test_vmuls_laneq_f32(float32_t a, float32x4_t b) {
- // CHECK-LABEL: test_vmuls_laneq_f32
return vmuls_laneq_f32(a, b, 3);
- // CHECK: fmul {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[3]
}
+// CHECK-LABEL: define double @test_vmuld_laneq_f64(double %a, <2 x double> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x double> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x double>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <2 x double> [[TMP1]], i32 1
+// CHECK: [[MUL:%.*]] = fmul double %a, [[VGETQ_LANE]]
+// CHECK: ret double [[MUL]]
float64_t test_vmuld_laneq_f64(float64_t a, float64x2_t b) {
- // CHECK-LABEL: test_vmuld_laneq_f64
return vmuld_laneq_f64(a, b, 1);
- // CHECK: fmul {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+}}.d[1]
}
+// CHECK-LABEL: define <1 x double> @test_vmul_n_f64(<1 x double> %a, double %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x double> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x double>
+// CHECK: [[TMP2:%.*]] = bitcast <1 x double> [[TMP1]] to double
+// CHECK: [[TMP3:%.*]] = fmul double [[TMP2]], %b
+// CHECK: [[TMP4:%.*]] = bitcast double [[TMP3]] to <1 x double>
+// CHECK: ret <1 x double> [[TMP4]]
float64x1_t test_vmul_n_f64(float64x1_t a, float64_t b) {
- // CHECK-LABEL: test_vmul_n_f64
return vmul_n_f64(a, b);
- // CHECK: fmul {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+.d\[0\]|d[0-9]+}}
}
+// CHECK-LABEL: define float @test_vmulxs_lane_f32(float %a, <2 x float> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <2 x float> [[TMP1]], i32 1
+// CHECK: [[VMULXS_F32_I:%.*]] = call float @llvm.aarch64.neon.fmulx.f32(float %a, float [[VGET_LANE]]) #2
+// CHECK: ret float [[VMULXS_F32_I]]
float32_t test_vmulxs_lane_f32(float32_t a, float32x2_t b) {
-// CHECK-LABEL: test_vmulxs_lane_f32
return vmulxs_lane_f32(a, b, 1);
-// CHECK: fmulx {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[1]
}
+// CHECK-LABEL: define float @test_vmulxs_laneq_f32(float %a, <4 x float> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <4 x float> [[TMP1]], i32 3
+// CHECK: [[VMULXS_F32_I:%.*]] = call float @llvm.aarch64.neon.fmulx.f32(float %a, float [[VGETQ_LANE]]) #2
+// CHECK: ret float [[VMULXS_F32_I]]
float32_t test_vmulxs_laneq_f32(float32_t a, float32x4_t b) {
-// CHECK-LABEL: test_vmulxs_laneq_f32
return vmulxs_laneq_f32(a, b, 3);
-// CHECK: fmulx {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[3]
}
+// CHECK-LABEL: define double @test_vmulxd_lane_f64(double %a, <1 x double> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x double> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x double>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <1 x double> [[TMP1]], i32 0
+// CHECK: [[VMULXD_F64_I:%.*]] = call double @llvm.aarch64.neon.fmulx.f64(double %a, double [[VGET_LANE]]) #2
+// CHECK: ret double [[VMULXD_F64_I]]
float64_t test_vmulxd_lane_f64(float64_t a, float64x1_t b) {
-// CHECK-LABEL: test_vmulxd_lane_f64
return vmulxd_lane_f64(a, b, 0);
-// CHECK: fmulx {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+.d\[0\]|d[0-9]+}}
}
+// CHECK-LABEL: define double @test_vmulxd_laneq_f64(double %a, <2 x double> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x double> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x double>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <2 x double> [[TMP1]], i32 1
+// CHECK: [[VMULXD_F64_I:%.*]] = call double @llvm.aarch64.neon.fmulx.f64(double %a, double [[VGETQ_LANE]]) #2
+// CHECK: ret double [[VMULXD_F64_I]]
float64_t test_vmulxd_laneq_f64(float64_t a, float64x2_t b) {
-// CHECK-LABEL: test_vmulxd_laneq_f64
return vmulxd_laneq_f64(a, b, 1);
-// CHECK: fmulx {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+}}.d[1]
}
-// CHECK-LABEL: test_vmulx_lane_f64
+// CHECK-LABEL: define <1 x double> @test_vmulx_lane_f64(<1 x double> %a, <1 x double> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x double> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x double>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <1 x double> [[TMP1]], i32 0
+// CHECK: [[TMP2:%.*]] = bitcast <1 x double> %b to <8 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP2]] to <1 x double>
+// CHECK: [[VGET_LANE6:%.*]] = extractelement <1 x double> [[TMP3]], i32 0
+// CHECK: [[VMULXD_F64_I:%.*]] = call double @llvm.aarch64.neon.fmulx.f64(double [[VGET_LANE]], double [[VGET_LANE]]6) #2
+// CHECK: [[TMP4:%.*]] = bitcast <1 x double> %a to <8 x i8>
+// CHECK: [[TMP5:%.*]] = bitcast <8 x i8> [[TMP4]] to <1 x double>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <1 x double> [[TMP5]], double [[VMULXD_F64_I]], i32 0
+// CHECK: ret <1 x double> [[VSET_LANE]]
float64x1_t test_vmulx_lane_f64(float64x1_t a, float64x1_t b) {
return vmulx_lane_f64(a, b, 0);
- // CHECK: fmulx {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+.d\[0\]|d[0-9]+}}
}
-// CHECK-LABEL: test_vmulx_laneq_f64_0
+// CHECK-LABEL: define <1 x double> @test_vmulx_laneq_f64_0(<1 x double> %a, <2 x double> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x double> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x double>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <1 x double> [[TMP1]], i32 0
+// CHECK: [[TMP2:%.*]] = bitcast <2 x double> %b to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP2]] to <2 x double>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <2 x double> [[TMP3]], i32 0
+// CHECK: [[VMULXD_F64_I:%.*]] = call double @llvm.aarch64.neon.fmulx.f64(double [[VGET_LANE]], double [[VGETQ_LANE]]) #2
+// CHECK: [[TMP4:%.*]] = bitcast <1 x double> %a to <8 x i8>
+// CHECK: [[TMP5:%.*]] = bitcast <8 x i8> [[TMP4]] to <1 x double>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <1 x double> [[TMP5]], double [[VMULXD_F64_I]], i32 0
+// CHECK: ret <1 x double> [[VSET_LANE]]
float64x1_t test_vmulx_laneq_f64_0(float64x1_t a, float64x2_t b) {
return vmulx_laneq_f64(a, b, 0);
- // CHECK: fmulx {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+}}.d[0]
}
-// CHECK-LABEL: test_vmulx_laneq_f64_1
+// CHECK-LABEL: define <1 x double> @test_vmulx_laneq_f64_1(<1 x double> %a, <2 x double> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x double> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x double>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <1 x double> [[TMP1]], i32 0
+// CHECK: [[TMP2:%.*]] = bitcast <2 x double> %b to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP2]] to <2 x double>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <2 x double> [[TMP3]], i32 1
+// CHECK: [[VMULXD_F64_I:%.*]] = call double @llvm.aarch64.neon.fmulx.f64(double [[VGET_LANE]], double [[VGETQ_LANE]]) #2
+// CHECK: [[TMP4:%.*]] = bitcast <1 x double> %a to <8 x i8>
+// CHECK: [[TMP5:%.*]] = bitcast <8 x i8> [[TMP4]] to <1 x double>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <1 x double> [[TMP5]], double [[VMULXD_F64_I]], i32 0
+// CHECK: ret <1 x double> [[VSET_LANE]]
float64x1_t test_vmulx_laneq_f64_1(float64x1_t a, float64x2_t b) {
return vmulx_laneq_f64(a, b, 1);
- // CHECK: fmulx {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+}}.d[1]
}
-// CHECK-LABEL: test_vfmas_lane_f32
+// CHECK-LABEL: define float @test_vfmas_lane_f32(float %a, float %b, <2 x float> %c) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %c to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[EXTRACT:%.*]] = extractelement <2 x float> [[TMP1]], i32 1
+// CHECK: [[TMP2:%.*]] = call float @llvm.fma.f32(float %b, float [[EXTRACT]], float %a)
+// CHECK: ret float [[TMP2]]
float32_t test_vfmas_lane_f32(float32_t a, float32_t b, float32x2_t c) {
return vfmas_lane_f32(a, b, c, 1);
- // CHECK: fmla {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[1]
}
-// CHECK-LABEL: test_vfmad_lane_f64
+// CHECK-LABEL: define double @test_vfmad_lane_f64(double %a, double %b, <1 x double> %c) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x double> %c to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x double>
+// CHECK: [[EXTRACT:%.*]] = extractelement <1 x double> [[TMP1]], i32 0
+// CHECK: [[TMP2:%.*]] = call double @llvm.fma.f64(double %b, double [[EXTRACT]], double %a)
+// CHECK: ret double [[TMP2]]
float64_t test_vfmad_lane_f64(float64_t a, float64_t b, float64x1_t c) {
return vfmad_lane_f64(a, b, c, 0);
- // CHECK: {{fmla|fmadd}} {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+.d\[0\]|d[0-9]+}}
}
-// CHECK-LABEL: test_vfmad_laneq_f64
+// CHECK-LABEL: define double @test_vfmad_laneq_f64(double %a, double %b, <2 x double> %c) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x double> %c to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x double>
+// CHECK: [[EXTRACT:%.*]] = extractelement <2 x double> [[TMP1]], i32 1
+// CHECK: [[TMP2:%.*]] = call double @llvm.fma.f64(double %b, double [[EXTRACT]], double %a)
+// CHECK: ret double [[TMP2]]
float64_t test_vfmad_laneq_f64(float64_t a, float64_t b, float64x2_t c) {
return vfmad_laneq_f64(a, b, c, 1);
- // CHECK: fmla {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+}}.d[1]
}
-// CHECK-LABEL: test_vfmss_lane_f32
+// CHECK-LABEL: define float @test_vfmss_lane_f32(float %a, float %b, <2 x float> %c) #0 {
+// CHECK: [[SUB:%.*]] = fsub <2 x float> <float -0.000000e+00, float -0.000000e+00>, %c
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> [[SUB]] to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[EXTRACT:%.*]] = extractelement <2 x float> [[TMP1]], i32 1
+// CHECK: [[TMP2:%.*]] = call float @llvm.fma.f32(float %b, float [[EXTRACT]], float %a)
+// CHECK: ret float [[TMP2]]
float32_t test_vfmss_lane_f32(float32_t a, float32_t b, float32x2_t c) {
return vfmss_lane_f32(a, b, c, 1);
- // CHECK: fmls {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[1]
}
-// CHECK-LABEL: test_vfma_lane_f64
+// CHECK-LABEL: define <1 x double> @test_vfma_lane_f64(<1 x double> %a, <1 x double> %b, <1 x double> %v) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x double> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <1 x double> %b to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <1 x double> %v to <8 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP2]] to <1 x double>
+// CHECK: [[LANE:%.*]] = shufflevector <1 x double> [[TMP3]], <1 x double> [[TMP3]], <1 x i32> zeroinitializer
+// CHECK: [[FMLA:%.*]] = bitcast <8 x i8> [[TMP1]] to <1 x double>
+// CHECK: [[FMLA1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x double>
+// CHECK: [[FMLA2:%.*]] = call <1 x double> @llvm.fma.v1f64(<1 x double> [[FMLA]], <1 x double> [[LANE]], <1 x double> [[FMLA]]1)
+// CHECK: ret <1 x double> [[FMLA]]2
float64x1_t test_vfma_lane_f64(float64x1_t a, float64x1_t b, float64x1_t v) {
return vfma_lane_f64(a, b, v, 0);
- // CHECK: {{fmla|fmadd}} {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+.d\[0\]|d[0-9]+}}
}
-// CHECK-LABEL: test_vfms_lane_f64
+// CHECK-LABEL: define <1 x double> @test_vfms_lane_f64(<1 x double> %a, <1 x double> %b, <1 x double> %v) #0 {
+// CHECK: [[SUB:%.*]] = fsub <1 x double> <double -0.000000e+00>, %v
+// CHECK: [[TMP0:%.*]] = bitcast <1 x double> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <1 x double> %b to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <1 x double> [[SUB]] to <8 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP2]] to <1 x double>
+// CHECK: [[LANE:%.*]] = shufflevector <1 x double> [[TMP3]], <1 x double> [[TMP3]], <1 x i32> zeroinitializer
+// CHECK: [[FMLA:%.*]] = bitcast <8 x i8> [[TMP1]] to <1 x double>
+// CHECK: [[FMLA1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x double>
+// CHECK: [[FMLA2:%.*]] = call <1 x double> @llvm.fma.v1f64(<1 x double> [[FMLA]], <1 x double> [[LANE]], <1 x double> [[FMLA]]1)
+// CHECK: ret <1 x double> [[FMLA]]2
float64x1_t test_vfms_lane_f64(float64x1_t a, float64x1_t b, float64x1_t v) {
return vfms_lane_f64(a, b, v, 0);
- // CHECK: {{fmls|fmsub}} {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+.d\[0\]|d[0-9]+}}
}
-// CHECK-LABEL: test_vfma_laneq_f64
+// CHECK-LABEL: define <1 x double> @test_vfma_laneq_f64(<1 x double> %a, <1 x double> %b, <2 x double> %v) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x double> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <1 x double> %b to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <2 x double> %v to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP0]] to double
+// CHECK: [[TMP4:%.*]] = bitcast <8 x i8> [[TMP1]] to double
+// CHECK: [[TMP5:%.*]] = bitcast <16 x i8> [[TMP2]] to <2 x double>
+// CHECK: [[EXTRACT:%.*]] = extractelement <2 x double> [[TMP5]], i32 0
+// CHECK: [[TMP6:%.*]] = call double @llvm.fma.f64(double [[TMP4]], double [[EXTRACT]], double [[TMP3]])
+// CHECK: [[TMP7:%.*]] = bitcast double [[TMP6]] to <1 x double>
+// CHECK: ret <1 x double> [[TMP7]]
float64x1_t test_vfma_laneq_f64(float64x1_t a, float64x1_t b, float64x2_t v) {
return vfma_laneq_f64(a, b, v, 0);
- // CHECK: fmla {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+}}.d[0]
}
-// CHECK-LABEL: test_vfms_laneq_f64
+// CHECK-LABEL: define <1 x double> @test_vfms_laneq_f64(<1 x double> %a, <1 x double> %b, <2 x double> %v) #0 {
+// CHECK: [[SUB:%.*]] = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, %v
+// CHECK: [[TMP0:%.*]] = bitcast <1 x double> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <1 x double> %b to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <2 x double> [[SUB]] to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP0]] to double
+// CHECK: [[TMP4:%.*]] = bitcast <8 x i8> [[TMP1]] to double
+// CHECK: [[TMP5:%.*]] = bitcast <16 x i8> [[TMP2]] to <2 x double>
+// CHECK: [[EXTRACT:%.*]] = extractelement <2 x double> [[TMP5]], i32 0
+// CHECK: [[TMP6:%.*]] = call double @llvm.fma.f64(double [[TMP4]], double [[EXTRACT]], double [[TMP3]])
+// CHECK: [[TMP7:%.*]] = bitcast double [[TMP6]] to <1 x double>
+// CHECK: ret <1 x double> [[TMP7]]
float64x1_t test_vfms_laneq_f64(float64x1_t a, float64x1_t b, float64x2_t v) {
return vfms_laneq_f64(a, b, v, 0);
- // CHECK: fmls {{d[0-9]+}}, {{d[0-9]+}}, {{v[0-9]+}}.d[0]
}
-// CHECK-LABEL: test_vqdmullh_lane_s16
+// CHECK-LABEL: define i32 @test_vqdmullh_lane_s16(i16 %a, <4 x i16> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <4 x i16> [[TMP1]], i32 3
+// CHECK: [[TMP2:%.*]] = insertelement <4 x i16> undef, i16 %a, i64 0
+// CHECK: [[TMP3:%.*]] = insertelement <4 x i16> undef, i16 [[VGET_LANE]], i64 0
+// CHECK: [[VQDMULLH_S16_I:%.*]] = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> [[TMP2]], <4 x i16> [[TMP3]]) #2
+// CHECK: [[TMP4:%.*]] = extractelement <4 x i32> [[VQDMULLH_S16_I]], i64 0
+// CHECK: ret i32 [[TMP4]]
int32_t test_vqdmullh_lane_s16(int16_t a, int16x4_t b) {
return vqdmullh_lane_s16(a, b, 3);
- // CHECK: sqdmull {{s[0-9]+|v[0-9]+.4s}}, {{h[0-9]+|v[0-9].4h}}, {{v[0-9]+}}.h[3]
}
-// CHECK-LABEL: test_vqdmulls_lane_s32
+// CHECK-LABEL: define i64 @test_vqdmulls_lane_s32(i32 %a, <2 x i32> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <2 x i32> [[TMP1]], i32 1
+// CHECK: [[VQDMULLS_S32_I:%.*]] = call i64 @llvm.aarch64.neon.sqdmulls.scalar(i32 %a, i32 [[VGET_LANE]]) #2
+// CHECK: ret i64 [[VQDMULLS_S32_I]]
int64_t test_vqdmulls_lane_s32(int32_t a, int32x2_t b) {
return vqdmulls_lane_s32(a, b, 1);
- // CHECK: sqdmull {{d[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[1]
}
-// CHECK-LABEL: test_vqdmullh_laneq_s16
+// CHECK-LABEL: define i32 @test_vqdmullh_laneq_s16(i16 %a, <8 x i16> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <8 x i16> [[TMP1]], i32 7
+// CHECK: [[TMP2:%.*]] = insertelement <4 x i16> undef, i16 %a, i64 0
+// CHECK: [[TMP3:%.*]] = insertelement <4 x i16> undef, i16 [[VGETQ_LANE]], i64 0
+// CHECK: [[VQDMULLH_S16_I:%.*]] = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> [[TMP2]], <4 x i16> [[TMP3]]) #2
+// CHECK: [[TMP4:%.*]] = extractelement <4 x i32> [[VQDMULLH_S16_I]], i64 0
+// CHECK: ret i32 [[TMP4]]
int32_t test_vqdmullh_laneq_s16(int16_t a, int16x8_t b) {
return vqdmullh_laneq_s16(a, b, 7);
- // CHECK: sqdmull {{s[0-9]+|v[0-9]+.4s}}, {{h[0-9]+|v[0-9]+.4h}}, {{v[0-9]+}}.h[7]
}
-// CHECK-LABEL: test_vqdmulls_laneq_s32
+// CHECK-LABEL: define i64 @test_vqdmulls_laneq_s32(i32 %a, <4 x i32> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <4 x i32> [[TMP1]], i32 3
+// CHECK: [[VQDMULLS_S32_I:%.*]] = call i64 @llvm.aarch64.neon.sqdmulls.scalar(i32 %a, i32 [[VGETQ_LANE]]) #2
+// CHECK: ret i64 [[VQDMULLS_S32_I]]
int64_t test_vqdmulls_laneq_s32(int32_t a, int32x4_t b) {
return vqdmulls_laneq_s32(a, b, 3);
- // CHECK: sqdmull {{d[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[3]
}
-// CHECK-LABEL: test_vqdmulhh_lane_s16
+// CHECK-LABEL: define i16 @test_vqdmulhh_lane_s16(i16 %a, <4 x i16> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <4 x i16> [[TMP1]], i32 3
+// CHECK: [[TMP2:%.*]] = insertelement <4 x i16> undef, i16 %a, i64 0
+// CHECK: [[TMP3:%.*]] = insertelement <4 x i16> undef, i16 [[VGET_LANE]], i64 0
+// CHECK: [[VQDMULHH_S16_I:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqdmulh.v4i16(<4 x i16> [[TMP2]], <4 x i16> [[TMP3]]) #2
+// CHECK: [[TMP4:%.*]] = extractelement <4 x i16> [[VQDMULHH_S16_I]], i64 0
+// CHECK: ret i16 [[TMP4]]
int16_t test_vqdmulhh_lane_s16(int16_t a, int16x4_t b) {
return vqdmulhh_lane_s16(a, b, 3);
-// CHECK: sqdmulh {{h[0-9]+|v[0-9]+.4h}}, {{h[0-9]+|v[0-9]+.4h}}, {{v[0-9]+}}.h[3]
}
-// CHECK-LABEL: test_vqdmulhs_lane_s32
+// CHECK-LABEL: define i32 @test_vqdmulhs_lane_s32(i32 %a, <2 x i32> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <2 x i32> [[TMP1]], i32 1
+// CHECK: [[VQDMULHS_S32_I:%.*]] = call i32 @llvm.aarch64.neon.sqdmulh.i32(i32 %a, i32 [[VGET_LANE]]) #2
+// CHECK: ret i32 [[VQDMULHS_S32_I]]
int32_t test_vqdmulhs_lane_s32(int32_t a, int32x2_t b) {
return vqdmulhs_lane_s32(a, b, 1);
-// CHECK: sqdmulh {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[1]
}
-// CHECK-LABEL: test_vqdmulhh_laneq_s16
+// CHECK-LABEL: define i16 @test_vqdmulhh_laneq_s16(i16 %a, <8 x i16> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <8 x i16> [[TMP1]], i32 7
+// CHECK: [[TMP2:%.*]] = insertelement <4 x i16> undef, i16 %a, i64 0
+// CHECK: [[TMP3:%.*]] = insertelement <4 x i16> undef, i16 [[VGETQ_LANE]], i64 0
+// CHECK: [[VQDMULHH_S16_I:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqdmulh.v4i16(<4 x i16> [[TMP2]], <4 x i16> [[TMP3]]) #2
+// CHECK: [[TMP4:%.*]] = extractelement <4 x i16> [[VQDMULHH_S16_I]], i64 0
+// CHECK: ret i16 [[TMP4]]
int16_t test_vqdmulhh_laneq_s16(int16_t a, int16x8_t b) {
return vqdmulhh_laneq_s16(a, b, 7);
-// CHECK: sqdmulh {{h[0-9]+|v[0-9]+.4h}}, {{h[0-9]+|v[0-9]+.4h}}, {{v[0-9]+}}.h[7]
}
-// CHECK-LABEL: test_vqdmulhs_laneq_s32
+// CHECK-LABEL: define i32 @test_vqdmulhs_laneq_s32(i32 %a, <4 x i32> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <4 x i32> [[TMP1]], i32 3
+// CHECK: [[VQDMULHS_S32_I:%.*]] = call i32 @llvm.aarch64.neon.sqdmulh.i32(i32 %a, i32 [[VGETQ_LANE]]) #2
+// CHECK: ret i32 [[VQDMULHS_S32_I]]
int32_t test_vqdmulhs_laneq_s32(int32_t a, int32x4_t b) {
return vqdmulhs_laneq_s32(a, b, 3);
-// CHECK: sqdmulh {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[3]
}
-// CHECK-LABEL: test_vqrdmulhh_lane_s16
+// CHECK-LABEL: define i16 @test_vqrdmulhh_lane_s16(i16 %a, <4 x i16> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <4 x i16> [[TMP1]], i32 3
+// CHECK: [[TMP2:%.*]] = insertelement <4 x i16> undef, i16 %a, i64 0
+// CHECK: [[TMP3:%.*]] = insertelement <4 x i16> undef, i16 [[VGET_LANE]], i64 0
+// CHECK: [[VQRDMULHH_S16_I:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqrdmulh.v4i16(<4 x i16> [[TMP2]], <4 x i16> [[TMP3]]) #2
+// CHECK: [[TMP4:%.*]] = extractelement <4 x i16> [[VQRDMULHH_S16_I]], i64 0
+// CHECK: ret i16 [[TMP4]]
int16_t test_vqrdmulhh_lane_s16(int16_t a, int16x4_t b) {
return vqrdmulhh_lane_s16(a, b, 3);
-// CHECK: sqrdmulh {{h[0-9]+|v[0-9]+.4h}}, {{h[0-9]+|v[0-9]+.4h}}, {{v[0-9]+}}.h[3]
}
-// CHECK-LABEL: test_vqrdmulhs_lane_s32
+// CHECK-LABEL: define i32 @test_vqrdmulhs_lane_s32(i32 %a, <2 x i32> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <2 x i32> [[TMP1]], i32 1
+// CHECK: [[VQRDMULHS_S32_I:%.*]] = call i32 @llvm.aarch64.neon.sqrdmulh.i32(i32 %a, i32 [[VGET_LANE]]) #2
+// CHECK: ret i32 [[VQRDMULHS_S32_I]]
int32_t test_vqrdmulhs_lane_s32(int32_t a, int32x2_t b) {
return vqrdmulhs_lane_s32(a, b, 1);
-// CHECK: sqrdmulh {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[1]
}
-// CHECK-LABEL: test_vqrdmulhh_laneq_s16
+// CHECK-LABEL: define i16 @test_vqrdmulhh_laneq_s16(i16 %a, <8 x i16> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <8 x i16> [[TMP1]], i32 7
+// CHECK: [[TMP2:%.*]] = insertelement <4 x i16> undef, i16 %a, i64 0
+// CHECK: [[TMP3:%.*]] = insertelement <4 x i16> undef, i16 [[VGETQ_LANE]], i64 0
+// CHECK: [[VQRDMULHH_S16_I:%.*]] = call <4 x i16> @llvm.aarch64.neon.sqrdmulh.v4i16(<4 x i16> [[TMP2]], <4 x i16> [[TMP3]]) #2
+// CHECK: [[TMP4:%.*]] = extractelement <4 x i16> [[VQRDMULHH_S16_I]], i64 0
+// CHECK: ret i16 [[TMP4]]
int16_t test_vqrdmulhh_laneq_s16(int16_t a, int16x8_t b) {
return vqrdmulhh_laneq_s16(a, b, 7);
-// CHECK: sqrdmulh {{h[0-9]+|v[0-9]+.4h}}, {{h[0-9]+|v[0-9]+.4h}}, {{v[0-9]+}}.h[7]
}
-// CHECK-LABEL: test_vqrdmulhs_laneq_s32
+// CHECK-LABEL: define i32 @test_vqrdmulhs_laneq_s32(i32 %a, <4 x i32> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <4 x i32> [[TMP1]], i32 3
+// CHECK: [[VQRDMULHS_S32_I:%.*]] = call i32 @llvm.aarch64.neon.sqrdmulh.i32(i32 %a, i32 [[VGETQ_LANE]]) #2
+// CHECK: ret i32 [[VQRDMULHS_S32_I]]
int32_t test_vqrdmulhs_laneq_s32(int32_t a, int32x4_t b) {
return vqrdmulhs_laneq_s32(a, b, 3);
-// CHECK: sqrdmulh {{s[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[3]
}
-// CHECK-LABEL: test_vqdmlalh_lane_s16
+// CHECK-LABEL: define i32 @test_vqdmlalh_lane_s16(i32 %a, i16 %b, <4 x i16> %c) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %c to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16>
+// CHECK: [[LANE:%.*]] = extractelement <4 x i16> [[TMP1]], i32 3
+// CHECK: [[TMP2:%.*]] = insertelement <4 x i16> undef, i16 %b, i64 0
+// CHECK: [[TMP3:%.*]] = insertelement <4 x i16> undef, i16 [[LANE]], i64 0
+// CHECK: [[VQDMLXL:%.*]] = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> [[TMP2]], <4 x i16> [[TMP3]])
+// CHECK: [[LANE0:%.*]] = extractelement <4 x i32> [[VQDMLXL]], i64 0
+// CHECK: [[VQDMLXL1:%.*]] = call i32 @llvm.aarch64.neon.sqadd.i32(i32 %a, i32 [[LANE]]0)
+// CHECK: ret i32 [[VQDMLXL]]1
int32_t test_vqdmlalh_lane_s16(int32_t a, int16_t b, int16x4_t c) {
return vqdmlalh_lane_s16(a, b, c, 3);
-// CHECK: sqdmlal {{s[0-9]+|v[0-9]+.4s}}, {{h[0-9]+|v[0-9]+.4h}}, {{v[0-9]+}}.h[3]
}
-// CHECK-LABEL: test_vqdmlals_lane_s32
+// CHECK-LABEL: define i64 @test_vqdmlals_lane_s32(i64 %a, i32 %b, <2 x i32> %c) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %c to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32>
+// CHECK: [[LANE:%.*]] = extractelement <2 x i32> [[TMP1]], i32 1
+// CHECK: [[VQDMLXL:%.*]] = call i64 @llvm.aarch64.neon.sqdmulls.scalar(i32 %b, i32 [[LANE]])
+// CHECK: [[VQDMLXL1:%.*]] = call i64 @llvm.aarch64.neon.sqadd.i64(i64 %a, i64 [[VQDMLXL]])
+// CHECK: ret i64 [[VQDMLXL]]1
int64_t test_vqdmlals_lane_s32(int64_t a, int32_t b, int32x2_t c) {
return vqdmlals_lane_s32(a, b, c, 1);
-// CHECK: sqdmlal {{d[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[1]
}
-// CHECK-LABEL: test_vqdmlalh_laneq_s16
+// CHECK-LABEL: define i32 @test_vqdmlalh_laneq_s16(i32 %a, i16 %b, <8 x i16> %c) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %c to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16>
+// CHECK: [[LANE:%.*]] = extractelement <8 x i16> [[TMP1]], i32 7
+// CHECK: [[TMP2:%.*]] = insertelement <4 x i16> undef, i16 %b, i64 0
+// CHECK: [[TMP3:%.*]] = insertelement <4 x i16> undef, i16 [[LANE]], i64 0
+// CHECK: [[VQDMLXL:%.*]] = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> [[TMP2]], <4 x i16> [[TMP3]])
+// CHECK: [[LANE0:%.*]] = extractelement <4 x i32> [[VQDMLXL]], i64 0
+// CHECK: [[VQDMLXL1:%.*]] = call i32 @llvm.aarch64.neon.sqadd.i32(i32 %a, i32 [[LANE]]0)
+// CHECK: ret i32 [[VQDMLXL]]1
int32_t test_vqdmlalh_laneq_s16(int32_t a, int16_t b, int16x8_t c) {
return vqdmlalh_laneq_s16(a, b, c, 7);
-// CHECK: sqdmlal {{s[0-9]+|v[0-9]+.4s}}, {{h[0-9]+|v[0-9]+.4h}}, {{v[0-9]+}}.h[7]
}
-// CHECK-LABEL: test_vqdmlals_laneq_s32
+// CHECK-LABEL: define i64 @test_vqdmlals_laneq_s32(i64 %a, i32 %b, <4 x i32> %c) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %c to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32>
+// CHECK: [[LANE:%.*]] = extractelement <4 x i32> [[TMP1]], i32 3
+// CHECK: [[VQDMLXL:%.*]] = call i64 @llvm.aarch64.neon.sqdmulls.scalar(i32 %b, i32 [[LANE]])
+// CHECK: [[VQDMLXL1:%.*]] = call i64 @llvm.aarch64.neon.sqadd.i64(i64 %a, i64 [[VQDMLXL]])
+// CHECK: ret i64 [[VQDMLXL]]1
int64_t test_vqdmlals_laneq_s32(int64_t a, int32_t b, int32x4_t c) {
return vqdmlals_laneq_s32(a, b, c, 3);
-// CHECK: sqdmlal {{d[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[3]
}
-// CHECK-LABEL: test_vqdmlslh_lane_s16
+// CHECK-LABEL: define i32 @test_vqdmlslh_lane_s16(i32 %a, i16 %b, <4 x i16> %c) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %c to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16>
+// CHECK: [[LANE:%.*]] = extractelement <4 x i16> [[TMP1]], i32 3
+// CHECK: [[TMP2:%.*]] = insertelement <4 x i16> undef, i16 %b, i64 0
+// CHECK: [[TMP3:%.*]] = insertelement <4 x i16> undef, i16 [[LANE]], i64 0
+// CHECK: [[VQDMLXL:%.*]] = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> [[TMP2]], <4 x i16> [[TMP3]])
+// CHECK: [[LANE0:%.*]] = extractelement <4 x i32> [[VQDMLXL]], i64 0
+// CHECK: [[VQDMLXL1:%.*]] = call i32 @llvm.aarch64.neon.sqsub.i32(i32 %a, i32 [[LANE]]0)
+// CHECK: ret i32 [[VQDMLXL]]1
int32_t test_vqdmlslh_lane_s16(int32_t a, int16_t b, int16x4_t c) {
return vqdmlslh_lane_s16(a, b, c, 3);
-// CHECK: sqdmlsl {{s[0-9]+|v[0-9]+.4s}}, {{h[0-9]+|v[0-9]+.4h}}, {{v[0-9]+}}.h[3]
}
-// CHECK-LABEL: test_vqdmlsls_lane_s32
+// CHECK-LABEL: define i64 @test_vqdmlsls_lane_s32(i64 %a, i32 %b, <2 x i32> %c) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %c to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32>
+// CHECK: [[LANE:%.*]] = extractelement <2 x i32> [[TMP1]], i32 1
+// CHECK: [[VQDMLXL:%.*]] = call i64 @llvm.aarch64.neon.sqdmulls.scalar(i32 %b, i32 [[LANE]])
+// CHECK: [[VQDMLXL1:%.*]] = call i64 @llvm.aarch64.neon.sqsub.i64(i64 %a, i64 [[VQDMLXL]])
+// CHECK: ret i64 [[VQDMLXL]]1
int64_t test_vqdmlsls_lane_s32(int64_t a, int32_t b, int32x2_t c) {
return vqdmlsls_lane_s32(a, b, c, 1);
-// CHECK: sqdmlsl {{d[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[1]
}
-// CHECK-LABEL: test_vqdmlslh_laneq_s16
+// CHECK-LABEL: define i32 @test_vqdmlslh_laneq_s16(i32 %a, i16 %b, <8 x i16> %c) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %c to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16>
+// CHECK: [[LANE:%.*]] = extractelement <8 x i16> [[TMP1]], i32 7
+// CHECK: [[TMP2:%.*]] = insertelement <4 x i16> undef, i16 %b, i64 0
+// CHECK: [[TMP3:%.*]] = insertelement <4 x i16> undef, i16 [[LANE]], i64 0
+// CHECK: [[VQDMLXL:%.*]] = call <4 x i32> @llvm.aarch64.neon.sqdmull.v4i32(<4 x i16> [[TMP2]], <4 x i16> [[TMP3]])
+// CHECK: [[LANE0:%.*]] = extractelement <4 x i32> [[VQDMLXL]], i64 0
+// CHECK: [[VQDMLXL1:%.*]] = call i32 @llvm.aarch64.neon.sqsub.i32(i32 %a, i32 [[LANE]]0)
+// CHECK: ret i32 [[VQDMLXL]]1
int32_t test_vqdmlslh_laneq_s16(int32_t a, int16_t b, int16x8_t c) {
return vqdmlslh_laneq_s16(a, b, c, 7);
-// CHECK: sqdmlsl {{s[0-9]+|v[0-9]+.4s}}, {{h[0-9]+|v[0-9]+.4h}}, {{v[0-9]+}}.h[7]
}
-// CHECK-LABEL: test_vqdmlsls_laneq_s32
+// CHECK-LABEL: define i64 @test_vqdmlsls_laneq_s32(i64 %a, i32 %b, <4 x i32> %c) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %c to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32>
+// CHECK: [[LANE:%.*]] = extractelement <4 x i32> [[TMP1]], i32 3
+// CHECK: [[VQDMLXL:%.*]] = call i64 @llvm.aarch64.neon.sqdmulls.scalar(i32 %b, i32 [[LANE]])
+// CHECK: [[VQDMLXL1:%.*]] = call i64 @llvm.aarch64.neon.sqsub.i64(i64 %a, i64 [[VQDMLXL]])
+// CHECK: ret i64 [[VQDMLXL]]1
int64_t test_vqdmlsls_laneq_s32(int64_t a, int32_t b, int32x4_t c) {
return vqdmlsls_laneq_s32(a, b, c, 3);
-// CHECK: sqdmlsl {{d[0-9]+}}, {{s[0-9]+}}, {{v[0-9]+}}.s[3]
}
-// CHECK-LABEL: test_vmulx_lane_f64_0:
+// CHECK-LABEL: define <1 x double> @test_vmulx_lane_f64_0() #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i64 4599917171378402754 to <1 x double>
+// CHECK: [[TMP1:%.*]] = bitcast i64 4606655882138939123 to <1 x double>
+// CHECK: [[TMP2:%.*]] = bitcast <1 x double> [[TMP0]] to <8 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP2]] to <1 x double>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <1 x double> [[TMP3]], i32 0
+// CHECK: [[TMP4:%.*]] = bitcast <1 x double> [[TMP1]] to <8 x i8>
+// CHECK: [[TMP5:%.*]] = bitcast <8 x i8> [[TMP4]] to <1 x double>
+// CHECK: [[VGET_LANE7:%.*]] = extractelement <1 x double> [[TMP5]], i32 0
+// CHECK: [[VMULXD_F64_I:%.*]] = call double @llvm.aarch64.neon.fmulx.f64(double [[VGET_LANE]], double [[VGET_LANE]]7) #2
+// CHECK: [[TMP6:%.*]] = bitcast <1 x double> [[TMP0]] to <8 x i8>
+// CHECK: [[TMP7:%.*]] = bitcast <8 x i8> [[TMP6]] to <1 x double>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <1 x double> [[TMP7]], double [[VMULXD_F64_I]], i32 0
+// CHECK: ret <1 x double> [[VSET_LANE]]
float64x1_t test_vmulx_lane_f64_0() {
float64x1_t arg1;
float64x1_t arg2;
@@ -262,15 +512,24 @@ float64x1_t test_vmulx_lane_f64_0() {
arg1 = vcreate_f64(UINT64_C(0x3fd6304bc43ab5c2));
arg2 = vcreate_f64(UINT64_C(0x3fee211e215aeef3));
result = vmulx_lane_f64(arg1, arg2, 0);
-// CHECK: adrp x[[ADDRLO:[0-9]+]]
-// CHECK: ldr d0, [x[[ADDRLO]],
-// CHECK: adrp x[[ADDRLO:[0-9]+]]
-// CHECK: ldr d1, [x[[ADDRLO]],
-// CHECK: fmulx d0, d1, d0
return result;
}
-// CHECK-LABEL: test_vmulx_laneq_f64_2:
+// CHECK-LABEL: define <1 x double> @test_vmulx_laneq_f64_2() #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i64 4599917171378402754 to <1 x double>
+// CHECK: [[TMP1:%.*]] = bitcast i64 4606655882138939123 to <1 x double>
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <1 x double> [[TMP0]], <1 x double> [[TMP1]], <2 x i32> <i32 0, i32 1>
+// CHECK: [[TMP2:%.*]] = bitcast <1 x double> [[TMP0]] to <8 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP2]] to <1 x double>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <1 x double> [[TMP3]], i32 0
+// CHECK: [[TMP4:%.*]] = bitcast <2 x double> [[SHUFFLE_I]] to <16 x i8>
+// CHECK: [[TMP5:%.*]] = bitcast <16 x i8> [[TMP4]] to <2 x double>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <2 x double> [[TMP5]], i32 1
+// CHECK: [[VMULXD_F64_I:%.*]] = call double @llvm.aarch64.neon.fmulx.f64(double [[VGET_LANE]], double [[VGETQ_LANE]]) #2
+// CHECK: [[TMP6:%.*]] = bitcast <1 x double> [[TMP0]] to <8 x i8>
+// CHECK: [[TMP7:%.*]] = bitcast <8 x i8> [[TMP6]] to <1 x double>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <1 x double> [[TMP7]], double [[VMULXD_F64_I]], i32 0
+// CHECK: ret <1 x double> [[VSET_LANE]]
float64x1_t test_vmulx_laneq_f64_2() {
float64x1_t arg1;
float64x1_t arg2;
@@ -281,10 +540,5 @@ float64x1_t test_vmulx_laneq_f64_2() {
arg2 = vcreate_f64(UINT64_C(0x3fee211e215aeef3));
arg3 = vcombine_f64(arg1, arg2);
result = vmulx_laneq_f64(arg1, arg3, 1);
-// CHECK: adrp x[[ADDRLO:[0-9]+]]
-// CHECK: ldr d0, [x[[ADDRLO]],
-// CHECK: adrp x[[ADDRLO:[0-9]+]]
-// CHECK: ldr d1, [x[[ADDRLO]],
-// CHECK: fmulx d0, d1, d0
return result;
}
Modified: cfe/trunk/test/CodeGen/aarch64-neon-shifts.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/aarch64-neon-shifts.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/aarch64-neon-shifts.c (original)
+++ cfe/trunk/test/CodeGen/aarch64-neon-shifts.c Wed Mar 9 12:54:42 2016
@@ -1,6 +1,5 @@
-// REQUIRES: aarch64-registered-target
// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon \
-// RUN: -ffp-contract=fast -S -emit-llvm -O1 -o - %s | FileCheck %s
+// RUN: -ffp-contract=fast -S -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s
#include <arm_neon.h>
@@ -25,19 +24,20 @@ uint8x8_t test_shift_vshr_umax(uint8x8_t
uint8x8_t test_shift_vsra(uint8x8_t a, uint8x8_t b) {
// CHECK-LABEL: test_shift_vsra
// CHECK: %[[SHR:.*]] = lshr <8 x i8> %b, <i8 5, i8 5, i8 5, i8 5, i8 5, i8 5, i8 5, i8 5>
- // CHECK: %{{.*}} = add <8 x i8> %[[SHR]], %a
+ // CHECK: %{{.*}} = add <8 x i8> %a, %[[SHR]]
return vsra_n_u8(a, b, 5);
}
int8x8_t test_shift_vsra_smax(int8x8_t a, int8x8_t b) {
// CHECK-LABEL: test_shift_vsra_smax
// CHECK: %[[SHR:.*]] = ashr <8 x i8> %b, <i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7>
- // CHECK: %{{.*}} = add <8 x i8> %[[SHR]], %a
+ // CHECK: %{{.*}} = add <8 x i8> %a, %[[SHR]]
return vsra_n_s8(a, b, 8);
}
uint8x8_t test_shift_vsra_umax(uint8x8_t a, uint8x8_t b) {
// CHECK-LABEL: test_shift_vsra_umax
- // CHECK: ret <8 x i8> %a
+ // CHECK: [[RES:%.*]] = add <8 x i8> %a, zeroinitializer
+ // CHECK: ret <8 x i8> [[RES]]
return vsra_n_u8(a, b, 8);
}
Modified: cfe/trunk/test/CodeGen/aarch64-neon-tbl.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/aarch64-neon-tbl.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/aarch64-neon-tbl.c (original)
+++ cfe/trunk/test/CodeGen/aarch64-neon-tbl.c Wed Mar 9 12:54:42 2016
@@ -1,463 +1,1500 @@
-// REQUIRES: aarch64-registered-target
// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon \
-// RUN: -ffp-contract=fast -S -O3 -o - %s | FileCheck %s
+// RUN: -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s
// Test new aarch64 intrinsics and types
#include <arm_neon.h>
+// CHECK-LABEL: define <8 x i8> @test_vtbl1_s8(<8 x i8> %a, <8 x i8> %b) #0 {
+// CHECK: [[VTBL1_I:%.*]] = shufflevector <8 x i8> %a, <8 x i8> zeroinitializer, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL11_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl1.v8i8(<16 x i8> [[VTBL1_I]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL11_I]]
int8x8_t test_vtbl1_s8(int8x8_t a, int8x8_t b) {
- // CHECK-LABEL: test_vtbl1_s8
return vtbl1_s8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbl1_s8(<16 x i8> %a, <8 x i8> %b) #0 {
+// CHECK: [[VTBL1_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl1.v8i8(<16 x i8> %a, <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL1_I]]
int8x8_t test_vqtbl1_s8(int8x16_t a, int8x8_t b) {
- // CHECK-LABEL: test_vqtbl1_s8
return vqtbl1_s8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbl2_s8([2 x <8 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.int8x8x2_t, align 8
+// CHECK: [[A:%.*]] = alloca %struct.int8x8x2_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x8x2_t, %struct.int8x8x2_t* [[A]], i32 0, i32 0
+// CHECK: store [2 x <8 x i8>] [[A]].coerce, [2 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x8x2_t, %struct.int8x8x2_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <8 x i8>], [2 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x8x2_t, %struct.int8x8x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [2 x <8 x i8>] [[TMP0]], [2 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x8x2_t, %struct.int8x8x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <8 x i8>], [2 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx.i, align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x8x2_t, %struct.int8x8x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <8 x i8>], [2 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx2.i, align 8
+// CHECK: [[VTBL1_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL13_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl1.v8i8(<16 x i8> [[VTBL1_I]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL13_I]]
int8x8_t test_vtbl2_s8(int8x8x2_t a, int8x8_t b) {
- // CHECK-LABEL: test_vtbl2_s8
return vtbl2_s8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbl2_s8([2 x <16 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.int8x16x2_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.int8x16x2_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[A]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[A]].coerce, [2 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <16 x i8>], [2 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[TMP0]], [2 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VTBL2_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl2.v8i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL2_I]]
int8x8_t test_vqtbl2_s8(int8x16x2_t a, int8x8_t b) {
- // CHECK-LABEL: test_vqtbl2_s8
return vqtbl2_s8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbl3_s8([3 x <8 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.int8x8x3_t, align 8
+// CHECK: [[A:%.*]] = alloca %struct.int8x8x3_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x8x3_t, %struct.int8x8x3_t* [[A]], i32 0, i32 0
+// CHECK: store [3 x <8 x i8>] [[A]].coerce, [3 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x8x3_t, %struct.int8x8x3_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <8 x i8>], [3 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x8x3_t, %struct.int8x8x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [3 x <8 x i8>] [[TMP0]], [3 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x8x3_t, %struct.int8x8x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx.i, align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x8x3_t, %struct.int8x8x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx2.i, align 8
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.int8x8x3_t, %struct.int8x8x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx4.i, align 8
+// CHECK: [[VTBL2_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL25_I:%.*]] = shufflevector <8 x i8> [[TMP3]], <8 x i8> zeroinitializer, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL26_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl2.v8i8(<16 x i8> [[VTBL2_I]], <16 x i8> [[VTBL25_I]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL26_I]]
int8x8_t test_vtbl3_s8(int8x8x3_t a, int8x8_t b) {
- // CHECK-LABEL: test_vtbl3_s8
return vtbl3_s8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbl3_s8([3 x <16 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.int8x16x3_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.int8x16x3_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[A]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[A]].coerce, [3 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <16 x i8>], [3 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[TMP0]], [3 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx4.i, align 16
+// CHECK: [[VTBL3_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl3.v8i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL3_I]]
int8x8_t test_vqtbl3_s8(int8x16x3_t a, int8x8_t b) {
- // CHECK-LABEL: test_vqtbl3_s8
return vqtbl3_s8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbl4_s8([4 x <8 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.int8x8x4_t, align 8
+// CHECK: [[A:%.*]] = alloca %struct.int8x8x4_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x8x4_t, %struct.int8x8x4_t* [[A]], i32 0, i32 0
+// CHECK: store [4 x <8 x i8>] [[A]].coerce, [4 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x8x4_t, %struct.int8x8x4_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <8 x i8>], [4 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x8x4_t, %struct.int8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [4 x <8 x i8>] [[TMP0]], [4 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x8x4_t, %struct.int8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx.i, align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x8x4_t, %struct.int8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx2.i, align 8
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.int8x8x4_t, %struct.int8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx4.i, align 8
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.int8x8x4_t, %struct.int8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx6.i, align 8
+// CHECK: [[VTBL2_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL27_I:%.*]] = shufflevector <8 x i8> [[TMP3]], <8 x i8> [[TMP4]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL28_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl2.v8i8(<16 x i8> [[VTBL2_I]], <16 x i8> [[VTBL27_I]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL28_I]]
int8x8_t test_vtbl4_s8(int8x8x4_t a, int8x8_t b) {
- // CHECK-LABEL: test_vtbl4_s8
return vtbl4_s8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbl4_s8([4 x <16 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.int8x16x4_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.int8x16x4_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[A]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[A]].coerce, [4 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <16 x i8>], [4 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[TMP0]], [4 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx4.i, align 16
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx6.i, align 16
+// CHECK: [[VTBL4_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl4.v8i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> [[TMP4]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL4_I]]
int8x8_t test_vqtbl4_s8(int8x16x4_t a, int8x8_t b) {
- // CHECK-LABEL: test_vqtbl4_s8
return vqtbl4_s8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbl1q_s8(<16 x i8> %a, <16 x i8> %b) #0 {
+// CHECK: [[VTBL1_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbl1.v16i8(<16 x i8> %a, <16 x i8> %b) #2
+// CHECK: ret <16 x i8> [[VTBL1_I]]
int8x16_t test_vqtbl1q_s8(int8x16_t a, int8x16_t b) {
- // CHECK-LABEL: test_vqtbl1q_s8
return vqtbl1q_s8(a, b);
- // CHECK: tbl {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbl2q_s8([2 x <16 x i8>] %a.coerce, <16 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.int8x16x2_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.int8x16x2_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[A]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[A]].coerce, [2 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <16 x i8>], [2 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[TMP0]], [2 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VTBL2_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbl2.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> %b) #2
+// CHECK: ret <16 x i8> [[VTBL2_I]]
int8x16_t test_vqtbl2q_s8(int8x16x2_t a, int8x16_t b) {
- // CHECK-LABEL: test_vqtbl2q_s8
return vqtbl2q_s8(a, b);
- // CHECK: tbl {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbl3q_s8([3 x <16 x i8>] %a.coerce, <16 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.int8x16x3_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.int8x16x3_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[A]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[A]].coerce, [3 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <16 x i8>], [3 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[TMP0]], [3 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx4.i, align 16
+// CHECK: [[VTBL3_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbl3.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> %b) #2
+// CHECK: ret <16 x i8> [[VTBL3_I]]
int8x16_t test_vqtbl3q_s8(int8x16x3_t a, int8x16_t b) {
- // CHECK-LABEL: test_vqtbl3q_s8
return vqtbl3q_s8(a, b);
- // CHECK: tbl {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbl4q_s8([4 x <16 x i8>] %a.coerce, <16 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.int8x16x4_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.int8x16x4_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[A]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[A]].coerce, [4 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <16 x i8>], [4 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[TMP0]], [4 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx4.i, align 16
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx6.i, align 16
+// CHECK: [[VTBL4_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbl4.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> [[TMP4]], <16 x i8> %b) #2
+// CHECK: ret <16 x i8> [[VTBL4_I]]
int8x16_t test_vqtbl4q_s8(int8x16x4_t a, int8x16_t b) {
- // CHECK-LABEL: test_vqtbl4q_s8
return vqtbl4q_s8(a, b);
- // CHECK: tbl {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbx1_s8(<8 x i8> %a, <8 x i8> %b, <8 x i8> %c) #0 {
+// CHECK: [[VTBL1_I:%.*]] = shufflevector <8 x i8> %b, <8 x i8> zeroinitializer, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL11_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl1.v8i8(<16 x i8> [[VTBL1_I]], <8 x i8> %c) #2
+// CHECK: [[TMP0:%.*]] = icmp uge <8 x i8> %c, <i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8>
+// CHECK: [[TMP1:%.*]] = sext <8 x i1> [[TMP0]] to <8 x i8>
+// CHECK: [[TMP2:%.*]] = and <8 x i8> [[TMP1]], %a
+// CHECK: [[TMP3:%.*]] = xor <8 x i8> [[TMP1]], <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
+// CHECK: [[TMP4:%.*]] = and <8 x i8> [[TMP3]], [[VTBL11_I]]
+// CHECK: [[VTBX_I:%.*]] = or <8 x i8> [[TMP2]], [[TMP4]]
+// CHECK: ret <8 x i8> [[VTBX_I]]
int8x8_t test_vtbx1_s8(int8x8_t a, int8x8_t b, int8x8_t c) {
- // CHECK-LABEL: test_vtbx1_s8
return vtbx1_s8(a, b, c);
- // CHECK: movi {{v[0-9]+.8b|d[0-9]+}}, #0
- // CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
- // CHECK: cmhs {{v[0-9]+}}.8b, {{v[0-9]+}}.8b, {{v[0-9]+}}.8b
- // CHECK: bsl {{v[0-9]+}}.8b, {{v[0-9]+}}.8b, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbx2_s8(<8 x i8> %a, [2 x <8 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.int8x8x2_t, align 8
+// CHECK: [[B:%.*]] = alloca %struct.int8x8x2_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x8x2_t, %struct.int8x8x2_t* [[B]], i32 0, i32 0
+// CHECK: store [2 x <8 x i8>] [[B]].coerce, [2 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x8x2_t, %struct.int8x8x2_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <8 x i8>], [2 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x8x2_t, %struct.int8x8x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [2 x <8 x i8>] [[TMP0]], [2 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x8x2_t, %struct.int8x8x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <8 x i8>], [2 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX_I]], align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x8x2_t, %struct.int8x8x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <8 x i8>], [2 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX2_I]], align 8
+// CHECK: [[VTBX1_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBX13_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx1.v8i8(<8 x i8> %a, <16 x i8> [[VTBX1_I]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX13_I]]
int8x8_t test_vtbx2_s8(int8x8_t a, int8x8x2_t b, int8x8_t c) {
- // CHECK-LABEL: test_vtbx2_s8
return vtbx2_s8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbx3_s8(<8 x i8> %a, [3 x <8 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.int8x8x3_t, align 8
+// CHECK: [[B:%.*]] = alloca %struct.int8x8x3_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x8x3_t, %struct.int8x8x3_t* [[B]], i32 0, i32 0
+// CHECK: store [3 x <8 x i8>] [[B]].coerce, [3 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x8x3_t, %struct.int8x8x3_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <8 x i8>], [3 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x8x3_t, %struct.int8x8x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [3 x <8 x i8>] [[TMP0]], [3 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x8x3_t, %struct.int8x8x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX_I]], align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x8x3_t, %struct.int8x8x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX2_I]], align 8
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.int8x8x3_t, %struct.int8x8x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX4_I]], align 8
+// CHECK: [[VTBL2_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL25_I:%.*]] = shufflevector <8 x i8> [[TMP3]], <8 x i8> zeroinitializer, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL26_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl2.v8i8(<16 x i8> [[VTBL2_I]], <16 x i8> [[VTBL25_I]], <8 x i8> %c) #2
+// CHECK: [[TMP4:%.*]] = icmp uge <8 x i8> %c, <i8 24, i8 24, i8 24, i8 24, i8 24, i8 24, i8 24, i8 24>
+// CHECK: [[TMP5:%.*]] = sext <8 x i1> [[TMP4]] to <8 x i8>
+// CHECK: [[TMP6:%.*]] = and <8 x i8> [[TMP5]], %a
+// CHECK: [[TMP7:%.*]] = xor <8 x i8> [[TMP5]], <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
+// CHECK: [[TMP8:%.*]] = and <8 x i8> [[TMP7]], [[VTBL26_I]]
+// CHECK: [[VTBX_I:%.*]] = or <8 x i8> [[TMP6]], [[TMP8]]
+// CHECK: ret <8 x i8> [[VTBX_I]]
int8x8_t test_vtbx3_s8(int8x8_t a, int8x8x3_t b, int8x8_t c) {
- // CHECK-LABEL: test_vtbx3_s8
return vtbx3_s8(a, b, c);
- // CHECK: movi {{v[0-9]+.8b|d[0-9]+}}, #0
- // CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
- // CHECK: cmhs {{v[0-9]+}}.8b, {{v[0-9]+}}.8b, {{v[0-9]+}}.8b
- // CHECK: bsl {{v[0-9]+}}.8b, {{v[0-9]+}}.8b, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbx4_s8(<8 x i8> %a, [4 x <8 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.int8x8x4_t, align 8
+// CHECK: [[B:%.*]] = alloca %struct.int8x8x4_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x8x4_t, %struct.int8x8x4_t* [[B]], i32 0, i32 0
+// CHECK: store [4 x <8 x i8>] [[B]].coerce, [4 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x8x4_t, %struct.int8x8x4_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <8 x i8>], [4 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x8x4_t, %struct.int8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [4 x <8 x i8>] [[TMP0]], [4 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x8x4_t, %struct.int8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX_I]], align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x8x4_t, %struct.int8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX2_I]], align 8
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.int8x8x4_t, %struct.int8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX4_I]], align 8
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.int8x8x4_t, %struct.int8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX6_I]], align 8
+// CHECK: [[VTBX2_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBX27_I:%.*]] = shufflevector <8 x i8> [[TMP3]], <8 x i8> [[TMP4]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBX28_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx2.v8i8(<8 x i8> %a, <16 x i8> [[VTBX2_I]], <16 x i8> [[VTBX27_I]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX28_I]]
int8x8_t test_vtbx4_s8(int8x8_t a, int8x8x4_t b, int8x8_t c) {
- // CHECK-LABEL: test_vtbx4_s8
return vtbx4_s8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbx1_s8(<8 x i8> %a, <16 x i8> %b, <8 x i8> %c) #0 {
+// CHECK: [[VTBX1_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx1.v8i8(<8 x i8> %a, <16 x i8> %b, <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX1_I]]
int8x8_t test_vqtbx1_s8(int8x8_t a, int8x16_t b, int8x8_t c) {
- // CHECK-LABEL: test_vqtbx1_s8
return vqtbx1_s8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbx2_s8(<8 x i8> %a, [2 x <16 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.int8x16x2_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.int8x16x2_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[B]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[B]].coerce, [2 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <16 x i8>], [2 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[TMP0]], [2 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VTBX2_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx2.v8i8(<8 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX2_I]]
int8x8_t test_vqtbx2_s8(int8x8_t a, int8x16x2_t b, int8x8_t c) {
- // CHECK-LABEL: test_vqtbx2_s8
return vqtbx2_s8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbx3_s8(<8 x i8> %a, [3 x <16 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.int8x16x3_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.int8x16x3_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[B]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[B]].coerce, [3 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <16 x i8>], [3 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[TMP0]], [3 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX4_I]], align 16
+// CHECK: [[VTBX3_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx3.v8i8(<8 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX3_I]]
int8x8_t test_vqtbx3_s8(int8x8_t a, int8x16x3_t b, int8x8_t c) {
- // CHECK-LABEL: test_vqtbx3_s8
return vqtbx3_s8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbx4_s8(<8 x i8> %a, [4 x <16 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.int8x16x4_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.int8x16x4_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[B]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[B]].coerce, [4 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <16 x i8>], [4 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[TMP0]], [4 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX4_I]], align 16
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX6_I]], align 16
+// CHECK: [[VTBX4_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx4.v8i8(<8 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> [[TMP4]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX4_I]]
int8x8_t test_vqtbx4_s8(int8x8_t a, int8x16x4_t b, int8x8_t c) {
- // CHECK-LABEL: test_vqtbx4_s8
return vqtbx4_s8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbx1q_s8(<16 x i8> %a, <16 x i8> %b, <16 x i8> %c) #0 {
+// CHECK: [[VTBX1_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbx1.v16i8(<16 x i8> %a, <16 x i8> %b, <16 x i8> %c) #2
+// CHECK: ret <16 x i8> [[VTBX1_I]]
int8x16_t test_vqtbx1q_s8(int8x16_t a, int8x16_t b, int8x16_t c) {
- // CHECK-LABEL: test_vqtbx1q_s8
return vqtbx1q_s8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbx2q_s8(<16 x i8> %a, [2 x <16 x i8>] %b.coerce, <16 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.int8x16x2_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.int8x16x2_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[B]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[B]].coerce, [2 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <16 x i8>], [2 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[TMP0]], [2 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x16x2_t, %struct.int8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VTBX2_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbx2.v16i8(<16 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> %c) #2
+// CHECK: ret <16 x i8> [[VTBX2_I]]
int8x16_t test_vqtbx2q_s8(int8x16_t a, int8x16x2_t b, int8x16_t c) {
- // CHECK-LABEL: test_vqtbx2q_s8
return vqtbx2q_s8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbx3q_s8(<16 x i8> %a, [3 x <16 x i8>] %b.coerce, <16 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.int8x16x3_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.int8x16x3_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[B]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[B]].coerce, [3 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <16 x i8>], [3 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[TMP0]], [3 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.int8x16x3_t, %struct.int8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX4_I]], align 16
+// CHECK: [[VTBX3_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbx3.v16i8(<16 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> %c) #2
+// CHECK: ret <16 x i8> [[VTBX3_I]]
int8x16_t test_vqtbx3q_s8(int8x16_t a, int8x16x3_t b, int8x16_t c) {
- // CHECK-LABEL: test_vqtbx3q_s8
return vqtbx3q_s8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbx4q_s8(<16 x i8> %a, [4 x <16 x i8>] %b.coerce, <16 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.int8x16x4_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.int8x16x4_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[B]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[B]].coerce, [4 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <16 x i8>], [4 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[TMP0]], [4 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX4_I]], align 16
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.int8x16x4_t, %struct.int8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX6_I]], align 16
+// CHECK: [[VTBX4_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbx4.v16i8(<16 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> [[TMP4]], <16 x i8> %c) #2
+// CHECK: ret <16 x i8> [[VTBX4_I]]
int8x16_t test_vqtbx4q_s8(int8x16_t a, int8x16x4_t b, int8x16_t c) {
- // CHECK-LABEL: test_vqtbx4q_s8
return vqtbx4q_s8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbl1_u8(<8 x i8> %a, <8 x i8> %b) #0 {
+// CHECK: [[VTBL1_I:%.*]] = shufflevector <8 x i8> %a, <8 x i8> zeroinitializer, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL11_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl1.v8i8(<16 x i8> [[VTBL1_I]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL11_I]]
uint8x8_t test_vtbl1_u8(uint8x8_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vtbl1_u8
return vtbl1_u8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbl1_u8(<16 x i8> %a, <8 x i8> %b) #0 {
+// CHECK: [[VTBL1_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl1.v8i8(<16 x i8> %a, <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL1_I]]
uint8x8_t test_vqtbl1_u8(uint8x16_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vqtbl1_u8
return vqtbl1_u8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbl2_u8([2 x <8 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.uint8x8x2_t, align 8
+// CHECK: [[A:%.*]] = alloca %struct.uint8x8x2_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x8x2_t, %struct.uint8x8x2_t* [[A]], i32 0, i32 0
+// CHECK: store [2 x <8 x i8>] [[A]].coerce, [2 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x8x2_t, %struct.uint8x8x2_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <8 x i8>], [2 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x8x2_t, %struct.uint8x8x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [2 x <8 x i8>] [[TMP0]], [2 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x8x2_t, %struct.uint8x8x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <8 x i8>], [2 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx.i, align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x8x2_t, %struct.uint8x8x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <8 x i8>], [2 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx2.i, align 8
+// CHECK: [[VTBL1_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL13_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl1.v8i8(<16 x i8> [[VTBL1_I]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL13_I]]
uint8x8_t test_vtbl2_u8(uint8x8x2_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vtbl2_u8
return vtbl2_u8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbl2_u8([2 x <16 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.uint8x16x2_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.uint8x16x2_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[A]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[A]].coerce, [2 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <16 x i8>], [2 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[TMP0]], [2 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VTBL2_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl2.v8i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL2_I]]
uint8x8_t test_vqtbl2_u8(uint8x16x2_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vqtbl2_u8
return vqtbl2_u8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbl3_u8([3 x <8 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.uint8x8x3_t, align 8
+// CHECK: [[A:%.*]] = alloca %struct.uint8x8x3_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x8x3_t, %struct.uint8x8x3_t* [[A]], i32 0, i32 0
+// CHECK: store [3 x <8 x i8>] [[A]].coerce, [3 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x8x3_t, %struct.uint8x8x3_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <8 x i8>], [3 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x8x3_t, %struct.uint8x8x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [3 x <8 x i8>] [[TMP0]], [3 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x8x3_t, %struct.uint8x8x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx.i, align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x8x3_t, %struct.uint8x8x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx2.i, align 8
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.uint8x8x3_t, %struct.uint8x8x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx4.i, align 8
+// CHECK: [[VTBL2_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL25_I:%.*]] = shufflevector <8 x i8> [[TMP3]], <8 x i8> zeroinitializer, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL26_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl2.v8i8(<16 x i8> [[VTBL2_I]], <16 x i8> [[VTBL25_I]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL26_I]]
uint8x8_t test_vtbl3_u8(uint8x8x3_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vtbl3_u8
return vtbl3_u8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbl3_u8([3 x <16 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.uint8x16x3_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.uint8x16x3_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[A]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[A]].coerce, [3 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <16 x i8>], [3 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[TMP0]], [3 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx4.i, align 16
+// CHECK: [[VTBL3_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl3.v8i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL3_I]]
uint8x8_t test_vqtbl3_u8(uint8x16x3_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vqtbl3_u8
return vqtbl3_u8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbl4_u8([4 x <8 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.uint8x8x4_t, align 8
+// CHECK: [[A:%.*]] = alloca %struct.uint8x8x4_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x8x4_t, %struct.uint8x8x4_t* [[A]], i32 0, i32 0
+// CHECK: store [4 x <8 x i8>] [[A]].coerce, [4 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x8x4_t, %struct.uint8x8x4_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <8 x i8>], [4 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x8x4_t, %struct.uint8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [4 x <8 x i8>] [[TMP0]], [4 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x8x4_t, %struct.uint8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx.i, align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x8x4_t, %struct.uint8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx2.i, align 8
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.uint8x8x4_t, %struct.uint8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx4.i, align 8
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.uint8x8x4_t, %struct.uint8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx6.i, align 8
+// CHECK: [[VTBL2_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL27_I:%.*]] = shufflevector <8 x i8> [[TMP3]], <8 x i8> [[TMP4]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL28_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl2.v8i8(<16 x i8> [[VTBL2_I]], <16 x i8> [[VTBL27_I]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL28_I]]
uint8x8_t test_vtbl4_u8(uint8x8x4_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vtbl4_u8
return vtbl4_u8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbl4_u8([4 x <16 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.uint8x16x4_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.uint8x16x4_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[A]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[A]].coerce, [4 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <16 x i8>], [4 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[TMP0]], [4 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx4.i, align 16
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx6.i, align 16
+// CHECK: [[VTBL4_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl4.v8i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> [[TMP4]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL4_I]]
uint8x8_t test_vqtbl4_u8(uint8x16x4_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vqtbl4_u8
return vqtbl4_u8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbl1q_u8(<16 x i8> %a, <16 x i8> %b) #0 {
+// CHECK: [[VTBL1_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbl1.v16i8(<16 x i8> %a, <16 x i8> %b) #2
+// CHECK: ret <16 x i8> [[VTBL1_I]]
uint8x16_t test_vqtbl1q_u8(uint8x16_t a, uint8x16_t b) {
- // CHECK-LABEL: test_vqtbl1q_u8
return vqtbl1q_u8(a, b);
- // CHECK: tbl {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbl2q_u8([2 x <16 x i8>] %a.coerce, <16 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.uint8x16x2_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.uint8x16x2_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[A]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[A]].coerce, [2 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <16 x i8>], [2 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[TMP0]], [2 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VTBL2_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbl2.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> %b) #2
+// CHECK: ret <16 x i8> [[VTBL2_I]]
uint8x16_t test_vqtbl2q_u8(uint8x16x2_t a, uint8x16_t b) {
- // CHECK-LABEL: test_vqtbl2q_u8
return vqtbl2q_u8(a, b);
- // CHECK: tbl {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbl3q_u8([3 x <16 x i8>] %a.coerce, <16 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.uint8x16x3_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.uint8x16x3_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[A]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[A]].coerce, [3 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <16 x i8>], [3 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[TMP0]], [3 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx4.i, align 16
+// CHECK: [[VTBL3_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbl3.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> %b) #2
+// CHECK: ret <16 x i8> [[VTBL3_I]]
uint8x16_t test_vqtbl3q_u8(uint8x16x3_t a, uint8x16_t b) {
- // CHECK-LABEL: test_vqtbl3q_u8
return vqtbl3q_u8(a, b);
- // CHECK: tbl {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbl4q_u8([4 x <16 x i8>] %a.coerce, <16 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.uint8x16x4_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.uint8x16x4_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[A]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[A]].coerce, [4 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <16 x i8>], [4 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[TMP0]], [4 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx4.i, align 16
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx6.i, align 16
+// CHECK: [[VTBL4_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbl4.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> [[TMP4]], <16 x i8> %b) #2
+// CHECK: ret <16 x i8> [[VTBL4_I]]
uint8x16_t test_vqtbl4q_u8(uint8x16x4_t a, uint8x16_t b) {
- // CHECK-LABEL: test_vqtbl4q_u8
return vqtbl4q_u8(a, b);
- // CHECK: tbl {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbx1_u8(<8 x i8> %a, <8 x i8> %b, <8 x i8> %c) #0 {
+// CHECK: [[VTBL1_I:%.*]] = shufflevector <8 x i8> %b, <8 x i8> zeroinitializer, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL11_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl1.v8i8(<16 x i8> [[VTBL1_I]], <8 x i8> %c) #2
+// CHECK: [[TMP0:%.*]] = icmp uge <8 x i8> %c, <i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8>
+// CHECK: [[TMP1:%.*]] = sext <8 x i1> [[TMP0]] to <8 x i8>
+// CHECK: [[TMP2:%.*]] = and <8 x i8> [[TMP1]], %a
+// CHECK: [[TMP3:%.*]] = xor <8 x i8> [[TMP1]], <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
+// CHECK: [[TMP4:%.*]] = and <8 x i8> [[TMP3]], [[VTBL11_I]]
+// CHECK: [[VTBX_I:%.*]] = or <8 x i8> [[TMP2]], [[TMP4]]
+// CHECK: ret <8 x i8> [[VTBX_I]]
uint8x8_t test_vtbx1_u8(uint8x8_t a, uint8x8_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vtbx1_u8
return vtbx1_u8(a, b, c);
- // CHECK: movi {{v[0-9]+.8b|d[0-9]+}}, #0
- // CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
- // CHECK: cmhs {{v[0-9]+}}.8b, {{v[0-9]+}}.8b, {{v[0-9]+}}.8b
- // CHECK: bsl {{v[0-9]+}}.8b, {{v[0-9]+}}.8b, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbx2_u8(<8 x i8> %a, [2 x <8 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.uint8x8x2_t, align 8
+// CHECK: [[B:%.*]] = alloca %struct.uint8x8x2_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x8x2_t, %struct.uint8x8x2_t* [[B]], i32 0, i32 0
+// CHECK: store [2 x <8 x i8>] [[B]].coerce, [2 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x8x2_t, %struct.uint8x8x2_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <8 x i8>], [2 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x8x2_t, %struct.uint8x8x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [2 x <8 x i8>] [[TMP0]], [2 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x8x2_t, %struct.uint8x8x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <8 x i8>], [2 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX_I]], align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x8x2_t, %struct.uint8x8x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <8 x i8>], [2 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX2_I]], align 8
+// CHECK: [[VTBX1_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBX13_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx1.v8i8(<8 x i8> %a, <16 x i8> [[VTBX1_I]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX13_I]]
uint8x8_t test_vtbx2_u8(uint8x8_t a, uint8x8x2_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vtbx2_u8
return vtbx2_u8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbx3_u8(<8 x i8> %a, [3 x <8 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.uint8x8x3_t, align 8
+// CHECK: [[B:%.*]] = alloca %struct.uint8x8x3_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x8x3_t, %struct.uint8x8x3_t* [[B]], i32 0, i32 0
+// CHECK: store [3 x <8 x i8>] [[B]].coerce, [3 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x8x3_t, %struct.uint8x8x3_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <8 x i8>], [3 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x8x3_t, %struct.uint8x8x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [3 x <8 x i8>] [[TMP0]], [3 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x8x3_t, %struct.uint8x8x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX_I]], align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x8x3_t, %struct.uint8x8x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX2_I]], align 8
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.uint8x8x3_t, %struct.uint8x8x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX4_I]], align 8
+// CHECK: [[VTBL2_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL25_I:%.*]] = shufflevector <8 x i8> [[TMP3]], <8 x i8> zeroinitializer, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL26_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl2.v8i8(<16 x i8> [[VTBL2_I]], <16 x i8> [[VTBL25_I]], <8 x i8> %c) #2
+// CHECK: [[TMP4:%.*]] = icmp uge <8 x i8> %c, <i8 24, i8 24, i8 24, i8 24, i8 24, i8 24, i8 24, i8 24>
+// CHECK: [[TMP5:%.*]] = sext <8 x i1> [[TMP4]] to <8 x i8>
+// CHECK: [[TMP6:%.*]] = and <8 x i8> [[TMP5]], %a
+// CHECK: [[TMP7:%.*]] = xor <8 x i8> [[TMP5]], <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
+// CHECK: [[TMP8:%.*]] = and <8 x i8> [[TMP7]], [[VTBL26_I]]
+// CHECK: [[VTBX_I:%.*]] = or <8 x i8> [[TMP6]], [[TMP8]]
+// CHECK: ret <8 x i8> [[VTBX_I]]
uint8x8_t test_vtbx3_u8(uint8x8_t a, uint8x8x3_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vtbx3_u8
return vtbx3_u8(a, b, c);
- // CHECK: movi {{v[0-9]+.8b|d[0-9]+}}, #0
- // CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
- // CHECK: cmhs {{v[0-9]+}}.8b, {{v[0-9]+}}.8b, {{v[0-9]+}}.8b
- // CHECK: bsl {{v[0-9]+}}.8b, {{v[0-9]+}}.8b, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbx4_u8(<8 x i8> %a, [4 x <8 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.uint8x8x4_t, align 8
+// CHECK: [[B:%.*]] = alloca %struct.uint8x8x4_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x8x4_t, %struct.uint8x8x4_t* [[B]], i32 0, i32 0
+// CHECK: store [4 x <8 x i8>] [[B]].coerce, [4 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x8x4_t, %struct.uint8x8x4_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <8 x i8>], [4 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x8x4_t, %struct.uint8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [4 x <8 x i8>] [[TMP0]], [4 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x8x4_t, %struct.uint8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX_I]], align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x8x4_t, %struct.uint8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX2_I]], align 8
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.uint8x8x4_t, %struct.uint8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX4_I]], align 8
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.uint8x8x4_t, %struct.uint8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX6_I]], align 8
+// CHECK: [[VTBX2_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBX27_I:%.*]] = shufflevector <8 x i8> [[TMP3]], <8 x i8> [[TMP4]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBX28_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx2.v8i8(<8 x i8> %a, <16 x i8> [[VTBX2_I]], <16 x i8> [[VTBX27_I]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX28_I]]
uint8x8_t test_vtbx4_u8(uint8x8_t a, uint8x8x4_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vtbx4_u8
return vtbx4_u8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbx1_u8(<8 x i8> %a, <16 x i8> %b, <8 x i8> %c) #0 {
+// CHECK: [[VTBX1_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx1.v8i8(<8 x i8> %a, <16 x i8> %b, <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX1_I]]
uint8x8_t test_vqtbx1_u8(uint8x8_t a, uint8x16_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vqtbx1_u8
return vqtbx1_u8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbx2_u8(<8 x i8> %a, [2 x <16 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.uint8x16x2_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.uint8x16x2_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[B]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[B]].coerce, [2 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <16 x i8>], [2 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[TMP0]], [2 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VTBX2_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx2.v8i8(<8 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX2_I]]
uint8x8_t test_vqtbx2_u8(uint8x8_t a, uint8x16x2_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vqtbx2_u8
return vqtbx2_u8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbx3_u8(<8 x i8> %a, [3 x <16 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.uint8x16x3_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.uint8x16x3_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[B]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[B]].coerce, [3 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <16 x i8>], [3 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[TMP0]], [3 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX4_I]], align 16
+// CHECK: [[VTBX3_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx3.v8i8(<8 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX3_I]]
uint8x8_t test_vqtbx3_u8(uint8x8_t a, uint8x16x3_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vqtbx3_u8
return vqtbx3_u8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbx4_u8(<8 x i8> %a, [4 x <16 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.uint8x16x4_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.uint8x16x4_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[B]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[B]].coerce, [4 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <16 x i8>], [4 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[TMP0]], [4 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX4_I]], align 16
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX6_I]], align 16
+// CHECK: [[VTBX4_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx4.v8i8(<8 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> [[TMP4]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX4_I]]
uint8x8_t test_vqtbx4_u8(uint8x8_t a, uint8x16x4_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vqtbx4_u8
return vqtbx4_u8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbx1q_u8(<16 x i8> %a, <16 x i8> %b, <16 x i8> %c) #0 {
+// CHECK: [[VTBX1_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbx1.v16i8(<16 x i8> %a, <16 x i8> %b, <16 x i8> %c) #2
+// CHECK: ret <16 x i8> [[VTBX1_I]]
uint8x16_t test_vqtbx1q_u8(uint8x16_t a, uint8x16_t b, uint8x16_t c) {
- // CHECK-LABEL: test_vqtbx1q_u8
return vqtbx1q_u8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbx2q_u8(<16 x i8> %a, [2 x <16 x i8>] %b.coerce, <16 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.uint8x16x2_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.uint8x16x2_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[B]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[B]].coerce, [2 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <16 x i8>], [2 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[TMP0]], [2 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x16x2_t, %struct.uint8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VTBX2_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbx2.v16i8(<16 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> %c) #2
+// CHECK: ret <16 x i8> [[VTBX2_I]]
uint8x16_t test_vqtbx2q_u8(uint8x16_t a, uint8x16x2_t b, uint8x16_t c) {
- // CHECK-LABEL: test_vqtbx2q_u8
return vqtbx2q_u8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbx3q_u8(<16 x i8> %a, [3 x <16 x i8>] %b.coerce, <16 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.uint8x16x3_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.uint8x16x3_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[B]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[B]].coerce, [3 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <16 x i8>], [3 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[TMP0]], [3 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.uint8x16x3_t, %struct.uint8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX4_I]], align 16
+// CHECK: [[VTBX3_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbx3.v16i8(<16 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> %c) #2
+// CHECK: ret <16 x i8> [[VTBX3_I]]
uint8x16_t test_vqtbx3q_u8(uint8x16_t a, uint8x16x3_t b, uint8x16_t c) {
- // CHECK-LABEL: test_vqtbx3q_u8
return vqtbx3q_u8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbx4q_u8(<16 x i8> %a, [4 x <16 x i8>] %b.coerce, <16 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.uint8x16x4_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.uint8x16x4_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[B]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[B]].coerce, [4 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <16 x i8>], [4 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[TMP0]], [4 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX4_I]], align 16
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.uint8x16x4_t, %struct.uint8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX6_I]], align 16
+// CHECK: [[VTBX4_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbx4.v16i8(<16 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> [[TMP4]], <16 x i8> %c) #2
+// CHECK: ret <16 x i8> [[VTBX4_I]]
uint8x16_t test_vqtbx4q_u8(uint8x16_t a, uint8x16x4_t b, uint8x16_t c) {
- // CHECK-LABEL: test_vqtbx4q_u8
return vqtbx4q_u8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbl1_p8(<8 x i8> %a, <8 x i8> %b) #0 {
+// CHECK: [[VTBL1_I:%.*]] = shufflevector <8 x i8> %a, <8 x i8> zeroinitializer, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL11_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl1.v8i8(<16 x i8> [[VTBL1_I]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL11_I]]
poly8x8_t test_vtbl1_p8(poly8x8_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vtbl1_p8
return vtbl1_p8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbl1_p8(<16 x i8> %a, <8 x i8> %b) #0 {
+// CHECK: [[VTBL1_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl1.v8i8(<16 x i8> %a, <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL1_I]]
poly8x8_t test_vqtbl1_p8(poly8x16_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vqtbl1_p8
return vqtbl1_p8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbl2_p8([2 x <8 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.poly8x8x2_t, align 8
+// CHECK: [[A:%.*]] = alloca %struct.poly8x8x2_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x8x2_t, %struct.poly8x8x2_t* [[A]], i32 0, i32 0
+// CHECK: store [2 x <8 x i8>] [[A]].coerce, [2 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x8x2_t, %struct.poly8x8x2_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <8 x i8>], [2 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x8x2_t, %struct.poly8x8x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [2 x <8 x i8>] [[TMP0]], [2 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x8x2_t, %struct.poly8x8x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <8 x i8>], [2 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx.i, align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x8x2_t, %struct.poly8x8x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <8 x i8>], [2 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx2.i, align 8
+// CHECK: [[VTBL1_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL13_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl1.v8i8(<16 x i8> [[VTBL1_I]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL13_I]]
poly8x8_t test_vtbl2_p8(poly8x8x2_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vtbl2_p8
return vtbl2_p8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbl2_p8([2 x <16 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.poly8x16x2_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.poly8x16x2_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[A]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[A]].coerce, [2 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <16 x i8>], [2 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[TMP0]], [2 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VTBL2_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl2.v8i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL2_I]]
poly8x8_t test_vqtbl2_p8(poly8x16x2_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vqtbl2_p8
return vqtbl2_p8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbl3_p8([3 x <8 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.poly8x8x3_t, align 8
+// CHECK: [[A:%.*]] = alloca %struct.poly8x8x3_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x8x3_t, %struct.poly8x8x3_t* [[A]], i32 0, i32 0
+// CHECK: store [3 x <8 x i8>] [[A]].coerce, [3 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x8x3_t, %struct.poly8x8x3_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <8 x i8>], [3 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x8x3_t, %struct.poly8x8x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [3 x <8 x i8>] [[TMP0]], [3 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x8x3_t, %struct.poly8x8x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx.i, align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x8x3_t, %struct.poly8x8x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx2.i, align 8
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.poly8x8x3_t, %struct.poly8x8x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx4.i, align 8
+// CHECK: [[VTBL2_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL25_I:%.*]] = shufflevector <8 x i8> [[TMP3]], <8 x i8> zeroinitializer, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL26_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl2.v8i8(<16 x i8> [[VTBL2_I]], <16 x i8> [[VTBL25_I]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL26_I]]
poly8x8_t test_vtbl3_p8(poly8x8x3_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vtbl3_p8
return vtbl3_p8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbl3_p8([3 x <16 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.poly8x16x3_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.poly8x16x3_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[A]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[A]].coerce, [3 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <16 x i8>], [3 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[TMP0]], [3 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx4.i, align 16
+// CHECK: [[VTBL3_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl3.v8i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL3_I]]
poly8x8_t test_vqtbl3_p8(poly8x16x3_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vqtbl3_p8
return vqtbl3_p8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbl4_p8([4 x <8 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.poly8x8x4_t, align 8
+// CHECK: [[A:%.*]] = alloca %struct.poly8x8x4_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x8x4_t, %struct.poly8x8x4_t* [[A]], i32 0, i32 0
+// CHECK: store [4 x <8 x i8>] [[A]].coerce, [4 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x8x4_t, %struct.poly8x8x4_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <8 x i8>], [4 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x8x4_t, %struct.poly8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [4 x <8 x i8>] [[TMP0]], [4 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x8x4_t, %struct.poly8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx.i, align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x8x4_t, %struct.poly8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx2.i, align 8
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.poly8x8x4_t, %struct.poly8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx4.i, align 8
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.poly8x8x4_t, %struct.poly8x8x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <8 x i8>, <8 x i8>* [[A]]rrayidx6.i, align 8
+// CHECK: [[VTBL2_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL27_I:%.*]] = shufflevector <8 x i8> [[TMP3]], <8 x i8> [[TMP4]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL28_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl2.v8i8(<16 x i8> [[VTBL2_I]], <16 x i8> [[VTBL27_I]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL28_I]]
poly8x8_t test_vtbl4_p8(poly8x8x4_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vtbl4_p8
return vtbl4_p8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbl4_p8([4 x <16 x i8>] %a.coerce, <8 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.poly8x16x4_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.poly8x16x4_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[A]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[A]].coerce, [4 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <16 x i8>], [4 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[TMP0]], [4 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx4.i, align 16
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx6.i, align 16
+// CHECK: [[VTBL4_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl4.v8i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> [[TMP4]], <8 x i8> %b) #2
+// CHECK: ret <8 x i8> [[VTBL4_I]]
poly8x8_t test_vqtbl4_p8(poly8x16x4_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vqtbl4_p8
return vqtbl4_p8(a, b);
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbl1q_p8(<16 x i8> %a, <16 x i8> %b) #0 {
+// CHECK: [[VTBL1_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbl1.v16i8(<16 x i8> %a, <16 x i8> %b) #2
+// CHECK: ret <16 x i8> [[VTBL1_I]]
poly8x16_t test_vqtbl1q_p8(poly8x16_t a, uint8x16_t b) {
- // CHECK-LABEL: test_vqtbl1q_p8
return vqtbl1q_p8(a, b);
- // CHECK: tbl {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbl2q_p8([2 x <16 x i8>] %a.coerce, <16 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.poly8x16x2_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.poly8x16x2_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[A]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[A]].coerce, [2 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <16 x i8>], [2 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[TMP0]], [2 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VTBL2_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbl2.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> %b) #2
+// CHECK: ret <16 x i8> [[VTBL2_I]]
poly8x16_t test_vqtbl2q_p8(poly8x16x2_t a, uint8x16_t b) {
- // CHECK-LABEL: test_vqtbl2q_p8
return vqtbl2q_p8(a, b);
- // CHECK: tbl {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbl3q_p8([3 x <16 x i8>] %a.coerce, <16 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.poly8x16x3_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.poly8x16x3_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[A]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[A]].coerce, [3 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <16 x i8>], [3 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[TMP0]], [3 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx4.i, align 16
+// CHECK: [[VTBL3_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbl3.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> %b) #2
+// CHECK: ret <16 x i8> [[VTBL3_I]]
poly8x16_t test_vqtbl3q_p8(poly8x16x3_t a, uint8x16_t b) {
- // CHECK-LABEL: test_vqtbl3q_p8
return vqtbl3q_p8(a, b);
- // CHECK: tbl {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbl4q_p8([4 x <16 x i8>] %a.coerce, <16 x i8> %b) #0 {
+// CHECK: [[__P0_I:%.*]] = alloca %struct.poly8x16x4_t, align 16
+// CHECK: [[A:%.*]] = alloca %struct.poly8x16x4_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[A]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[A]].coerce, [4 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[A]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <16 x i8>], [4 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[TMP0]], [4 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx.i, align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx2.i, align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx4.i, align 16
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P0_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* [[A]]rrayidx6.i, align 16
+// CHECK: [[VTBL4_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbl4.v16i8(<16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> [[TMP4]], <16 x i8> %b) #2
+// CHECK: ret <16 x i8> [[VTBL4_I]]
poly8x16_t test_vqtbl4q_p8(poly8x16x4_t a, uint8x16_t b) {
- // CHECK-LABEL: test_vqtbl4q_p8
return vqtbl4q_p8(a, b);
- // CHECK: tbl {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbx1_p8(<8 x i8> %a, <8 x i8> %b, <8 x i8> %c) #0 {
+// CHECK: [[VTBL1_I:%.*]] = shufflevector <8 x i8> %b, <8 x i8> zeroinitializer, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL11_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl1.v8i8(<16 x i8> [[VTBL1_I]], <8 x i8> %c) #2
+// CHECK: [[TMP0:%.*]] = icmp uge <8 x i8> %c, <i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8, i8 8>
+// CHECK: [[TMP1:%.*]] = sext <8 x i1> [[TMP0]] to <8 x i8>
+// CHECK: [[TMP2:%.*]] = and <8 x i8> [[TMP1]], %a
+// CHECK: [[TMP3:%.*]] = xor <8 x i8> [[TMP1]], <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
+// CHECK: [[TMP4:%.*]] = and <8 x i8> [[TMP3]], [[VTBL11_I]]
+// CHECK: [[VTBX_I:%.*]] = or <8 x i8> [[TMP2]], [[TMP4]]
+// CHECK: ret <8 x i8> [[VTBX_I]]
poly8x8_t test_vtbx1_p8(poly8x8_t a, poly8x8_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vtbx1_p8
return vtbx1_p8(a, b, c);
- // CHECK: movi {{v[0-9]+.8b|d[0-9]+}}, #0
- // CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
- // CHECK: cmhs {{v[0-9]+}}.8b, {{v[0-9]+}}.8b, {{v[0-9]+}}.8b
- // CHECK: bsl {{v[0-9]+}}.8b, {{v[0-9]+}}.8b, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbx2_p8(<8 x i8> %a, [2 x <8 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.poly8x8x2_t, align 8
+// CHECK: [[B:%.*]] = alloca %struct.poly8x8x2_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x8x2_t, %struct.poly8x8x2_t* [[B]], i32 0, i32 0
+// CHECK: store [2 x <8 x i8>] [[B]].coerce, [2 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x8x2_t, %struct.poly8x8x2_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <8 x i8>], [2 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x8x2_t, %struct.poly8x8x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [2 x <8 x i8>] [[TMP0]], [2 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x8x2_t, %struct.poly8x8x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <8 x i8>], [2 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX_I]], align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x8x2_t, %struct.poly8x8x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <8 x i8>], [2 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX2_I]], align 8
+// CHECK: [[VTBX1_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBX13_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx1.v8i8(<8 x i8> %a, <16 x i8> [[VTBX1_I]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX13_I]]
poly8x8_t test_vtbx2_p8(poly8x8_t a, poly8x8x2_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vtbx2_p8
return vtbx2_p8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbx3_p8(<8 x i8> %a, [3 x <8 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.poly8x8x3_t, align 8
+// CHECK: [[B:%.*]] = alloca %struct.poly8x8x3_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x8x3_t, %struct.poly8x8x3_t* [[B]], i32 0, i32 0
+// CHECK: store [3 x <8 x i8>] [[B]].coerce, [3 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x8x3_t, %struct.poly8x8x3_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <8 x i8>], [3 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x8x3_t, %struct.poly8x8x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [3 x <8 x i8>] [[TMP0]], [3 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x8x3_t, %struct.poly8x8x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX_I]], align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x8x3_t, %struct.poly8x8x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX2_I]], align 8
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.poly8x8x3_t, %struct.poly8x8x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <8 x i8>], [3 x <8 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX4_I]], align 8
+// CHECK: [[VTBL2_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL25_I:%.*]] = shufflevector <8 x i8> [[TMP3]], <8 x i8> zeroinitializer, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBL26_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbl2.v8i8(<16 x i8> [[VTBL2_I]], <16 x i8> [[VTBL25_I]], <8 x i8> %c) #2
+// CHECK: [[TMP4:%.*]] = icmp uge <8 x i8> %c, <i8 24, i8 24, i8 24, i8 24, i8 24, i8 24, i8 24, i8 24>
+// CHECK: [[TMP5:%.*]] = sext <8 x i1> [[TMP4]] to <8 x i8>
+// CHECK: [[TMP6:%.*]] = and <8 x i8> [[TMP5]], %a
+// CHECK: [[TMP7:%.*]] = xor <8 x i8> [[TMP5]], <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
+// CHECK: [[TMP8:%.*]] = and <8 x i8> [[TMP7]], [[VTBL26_I]]
+// CHECK: [[VTBX_I:%.*]] = or <8 x i8> [[TMP6]], [[TMP8]]
+// CHECK: ret <8 x i8> [[VTBX_I]]
poly8x8_t test_vtbx3_p8(poly8x8_t a, poly8x8x3_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vtbx3_p8
return vtbx3_p8(a, b, c);
- // CHECK: movi {{v[0-9]+.8b|d[0-9]+}}, #0
- // CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
- // CHECK: tbl {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
- // CHECK: cmhs {{v[0-9]+}}.8b, {{v[0-9]+}}.8b, {{v[0-9]+}}.8b
- // CHECK: bsl {{v[0-9]+}}.8b, {{v[0-9]+}}.8b, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vtbx4_p8(<8 x i8> %a, [4 x <8 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.poly8x8x4_t, align 8
+// CHECK: [[B:%.*]] = alloca %struct.poly8x8x4_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x8x4_t, %struct.poly8x8x4_t* [[B]], i32 0, i32 0
+// CHECK: store [4 x <8 x i8>] [[B]].coerce, [4 x <8 x i8>]* [[COERCE_DIVE]], align 8
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x8x4_t, %struct.poly8x8x4_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <8 x i8>], [4 x <8 x i8>]* [[COERCE_DIVE]]1, align 8
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x8x4_t, %struct.poly8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [4 x <8 x i8>] [[TMP0]], [4 x <8 x i8>]* [[COERCE_DIVE]].i, align 8
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x8x4_t, %struct.poly8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX_I]], align 8
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x8x4_t, %struct.poly8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX2_I]], align 8
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.poly8x8x4_t, %struct.poly8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX4_I]], align 8
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.poly8x8x4_t, %struct.poly8x8x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <8 x i8>], [4 x <8 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <8 x i8>, <8 x i8>* [[ARRAYIDX6_I]], align 8
+// CHECK: [[VTBX2_I:%.*]] = shufflevector <8 x i8> [[TMP1]], <8 x i8> [[TMP2]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBX27_I:%.*]] = shufflevector <8 x i8> [[TMP3]], <8 x i8> [[TMP4]], <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: [[VTBX28_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx2.v8i8(<8 x i8> %a, <16 x i8> [[VTBX2_I]], <16 x i8> [[VTBX27_I]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX28_I]]
poly8x8_t test_vtbx4_p8(poly8x8_t a, poly8x8x4_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vtbx4_p8
return vtbx4_p8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbx1_p8(<8 x i8> %a, <16 x i8> %b, <8 x i8> %c) #0 {
+// CHECK: [[VTBX1_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx1.v8i8(<8 x i8> %a, <16 x i8> %b, <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX1_I]]
poly8x8_t test_vqtbx1_p8(poly8x8_t a, uint8x16_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vqtbx1_p8
return vqtbx1_p8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbx2_p8(<8 x i8> %a, [2 x <16 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.poly8x16x2_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.poly8x16x2_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[B]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[B]].coerce, [2 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <16 x i8>], [2 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[TMP0]], [2 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VTBX2_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx2.v8i8(<8 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX2_I]]
poly8x8_t test_vqtbx2_p8(poly8x8_t a, poly8x16x2_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vqtbx2_p8
return vqtbx2_p8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbx3_p8(<8 x i8> %a, [3 x <16 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.poly8x16x3_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.poly8x16x3_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[B]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[B]].coerce, [3 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <16 x i8>], [3 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[TMP0]], [3 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX4_I]], align 16
+// CHECK: [[VTBX3_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx3.v8i8(<8 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX3_I]]
poly8x8_t test_vqtbx3_p8(poly8x8_t a, poly8x16x3_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vqtbx3_p8
return vqtbx3_p8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <8 x i8> @test_vqtbx4_p8(<8 x i8> %a, [4 x <16 x i8>] %b.coerce, <8 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.poly8x16x4_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.poly8x16x4_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[B]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[B]].coerce, [4 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <16 x i8>], [4 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[TMP0]], [4 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX4_I]], align 16
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX6_I]], align 16
+// CHECK: [[VTBX4_I:%.*]] = call <8 x i8> @llvm.aarch64.neon.tbx4.v8i8(<8 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> [[TMP4]], <8 x i8> %c) #2
+// CHECK: ret <8 x i8> [[VTBX4_I]]
poly8x8_t test_vqtbx4_p8(poly8x8_t a, poly8x16x4_t b, uint8x8_t c) {
- // CHECK-LABEL: test_vqtbx4_p8
return vqtbx4_p8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.8b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbx1q_p8(<16 x i8> %a, <16 x i8> %b, <16 x i8> %c) #0 {
+// CHECK: [[VTBX1_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbx1.v16i8(<16 x i8> %a, <16 x i8> %b, <16 x i8> %c) #2
+// CHECK: ret <16 x i8> [[VTBX1_I]]
poly8x16_t test_vqtbx1q_p8(poly8x16_t a, uint8x16_t b, uint8x16_t c) {
- // CHECK-LABEL: test_vqtbx1q_p8
return vqtbx1q_p8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbx2q_p8(<16 x i8> %a, [2 x <16 x i8>] %b.coerce, <16 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.poly8x16x2_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.poly8x16x2_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[B]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[B]].coerce, [2 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [2 x <16 x i8>], [2 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [2 x <16 x i8>] [[TMP0]], [2 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x16x2_t, %struct.poly8x16x2_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [2 x <16 x i8>], [2 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VTBX2_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbx2.v16i8(<16 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> %c) #2
+// CHECK: ret <16 x i8> [[VTBX2_I]]
poly8x16_t test_vqtbx2q_p8(poly8x16_t a, poly8x16x2_t b, uint8x16_t c) {
- // CHECK-LABEL: test_vqtbx2q_p8
return vqtbx2q_p8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbx3q_p8(<16 x i8> %a, [3 x <16 x i8>] %b.coerce, <16 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.poly8x16x3_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.poly8x16x3_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[B]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[B]].coerce, [3 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [3 x <16 x i8>], [3 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [3 x <16 x i8>] [[TMP0]], [3 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.poly8x16x3_t, %struct.poly8x16x3_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [3 x <16 x i8>], [3 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX4_I]], align 16
+// CHECK: [[VTBX3_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbx3.v16i8(<16 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> %c) #2
+// CHECK: ret <16 x i8> [[VTBX3_I]]
poly8x16_t test_vqtbx3q_p8(poly8x16_t a, poly8x16x3_t b, uint8x16_t c) {
- // CHECK-LABEL: test_vqtbx3q_p8
return vqtbx3q_p8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define <16 x i8> @test_vqtbx4q_p8(<16 x i8> %a, [4 x <16 x i8>] %b.coerce, <16 x i8> %c) #0 {
+// CHECK: [[__P1_I:%.*]] = alloca %struct.poly8x16x4_t, align 16
+// CHECK: [[B:%.*]] = alloca %struct.poly8x16x4_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[B]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[B]].coerce, [4 x <16 x i8>]* [[COERCE_DIVE]], align 16
+// CHECK: [[COERCE_DIVE1:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[B]], i32 0, i32 0
+// CHECK: [[TMP0:%.*]] = load [4 x <16 x i8>], [4 x <16 x i8>]* [[COERCE_DIVE]]1, align 16
+// CHECK: [[COERCE_DIVE_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: store [4 x <16 x i8>] [[TMP0]], [4 x <16 x i8>]* [[COERCE_DIVE]].i, align 16
+// CHECK: [[VAL_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL_I]], i64 0, i64 0
+// CHECK: [[TMP1:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX_I]], align 16
+// CHECK: [[VAL1_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL1_I]], i64 0, i64 1
+// CHECK: [[TMP2:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX2_I]], align 16
+// CHECK: [[VAL3_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX4_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL3_I]], i64 0, i64 2
+// CHECK: [[TMP3:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX4_I]], align 16
+// CHECK: [[VAL5_I:%.*]] = getelementptr inbounds %struct.poly8x16x4_t, %struct.poly8x16x4_t* [[__P1_I]], i32 0, i32 0
+// CHECK: [[ARRAYIDX6_I:%.*]] = getelementptr inbounds [4 x <16 x i8>], [4 x <16 x i8>]* [[VAL5_I]], i64 0, i64 3
+// CHECK: [[TMP4:%.*]] = load <16 x i8>, <16 x i8>* [[ARRAYIDX6_I]], align 16
+// CHECK: [[VTBX4_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.tbx4.v16i8(<16 x i8> %a, <16 x i8> [[TMP1]], <16 x i8> [[TMP2]], <16 x i8> [[TMP3]], <16 x i8> [[TMP4]], <16 x i8> %c) #2
+// CHECK: ret <16 x i8> [[VTBX4_I]]
poly8x16_t test_vqtbx4q_p8(poly8x16_t a, poly8x16x4_t b, uint8x16_t c) {
- // CHECK-LABEL: test_vqtbx4q_p8
return vqtbx4q_p8(a, b, c);
- // CHECK: tbx {{v[0-9]+}}.16b, {{{ ?v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b, v[0-9]+.16b ?}}}, {{v[0-9]+}}.16b
}
Modified: cfe/trunk/test/CodeGen/aarch64-neon-vcombine.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/aarch64-neon-vcombine.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/aarch64-neon-vcombine.c (original)
+++ cfe/trunk/test/CodeGen/aarch64-neon-vcombine.c Wed Mar 9 12:54:42 2016
@@ -1,90 +1,103 @@
-// REQUIRES: aarch64-registered-target
-// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon -S -O3 -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon -fallow-half-arguments-and-returns -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s
// Test new aarch64 intrinsics and types
#include <arm_neon.h>
+// CHECK-LABEL: define <16 x i8> @test_vcombine_s8(<8 x i8> %low, <8 x i8> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x i8> %low, <8 x i8> %high, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: ret <16 x i8> [[SHUFFLE_I]]
int8x16_t test_vcombine_s8(int8x8_t low, int8x8_t high) {
- // CHECK-LABEL: test_vcombine_s8:
return vcombine_s8(low, high);
- // CHECK: ins v0.d[1], v1.d[0]
}
+// CHECK-LABEL: define <8 x i16> @test_vcombine_s16(<4 x i16> %low, <4 x i16> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <4 x i16> %low, <4 x i16> %high, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+// CHECK: ret <8 x i16> [[SHUFFLE_I]]
int16x8_t test_vcombine_s16(int16x4_t low, int16x4_t high) {
- // CHECK-LABEL: test_vcombine_s16:
return vcombine_s16(low, high);
- // CHECK: ins v0.d[1], v1.d[0]
}
+// CHECK-LABEL: define <4 x i32> @test_vcombine_s32(<2 x i32> %low, <2 x i32> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x i32> %low, <2 x i32> %high, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+// CHECK: ret <4 x i32> [[SHUFFLE_I]]
int32x4_t test_vcombine_s32(int32x2_t low, int32x2_t high) {
- // CHECK-LABEL: test_vcombine_s32:
return vcombine_s32(low, high);
- // CHECK: ins v0.d[1], v1.d[0]
}
+// CHECK-LABEL: define <2 x i64> @test_vcombine_s64(<1 x i64> %low, <1 x i64> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <1 x i64> %low, <1 x i64> %high, <2 x i32> <i32 0, i32 1>
+// CHECK: ret <2 x i64> [[SHUFFLE_I]]
int64x2_t test_vcombine_s64(int64x1_t low, int64x1_t high) {
- // CHECK-LABEL: test_vcombine_s64:
return vcombine_s64(low, high);
- // CHECK: ins v0.d[1], v1.d[0]
}
+// CHECK-LABEL: define <16 x i8> @test_vcombine_u8(<8 x i8> %low, <8 x i8> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x i8> %low, <8 x i8> %high, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: ret <16 x i8> [[SHUFFLE_I]]
uint8x16_t test_vcombine_u8(uint8x8_t low, uint8x8_t high) {
- // CHECK-LABEL: test_vcombine_u8:
return vcombine_u8(low, high);
- // CHECK: ins v0.d[1], v1.d[0]
}
+// CHECK-LABEL: define <8 x i16> @test_vcombine_u16(<4 x i16> %low, <4 x i16> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <4 x i16> %low, <4 x i16> %high, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+// CHECK: ret <8 x i16> [[SHUFFLE_I]]
uint16x8_t test_vcombine_u16(uint16x4_t low, uint16x4_t high) {
- // CHECK-LABEL: test_vcombine_u16:
return vcombine_u16(low, high);
- // CHECK: ins v0.d[1], v1.d[0]
}
+// CHECK-LABEL: define <4 x i32> @test_vcombine_u32(<2 x i32> %low, <2 x i32> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x i32> %low, <2 x i32> %high, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+// CHECK: ret <4 x i32> [[SHUFFLE_I]]
uint32x4_t test_vcombine_u32(uint32x2_t low, uint32x2_t high) {
- // CHECK-LABEL: test_vcombine_u32:
return vcombine_u32(low, high);
- // CHECK: ins v0.d[1], v1.d[0]
}
+// CHECK-LABEL: define <2 x i64> @test_vcombine_u64(<1 x i64> %low, <1 x i64> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <1 x i64> %low, <1 x i64> %high, <2 x i32> <i32 0, i32 1>
+// CHECK: ret <2 x i64> [[SHUFFLE_I]]
uint64x2_t test_vcombine_u64(uint64x1_t low, uint64x1_t high) {
- // CHECK-LABEL: test_vcombine_u64:
return vcombine_u64(low, high);
- // CHECK: ins v0.d[1], v1.d[0]
}
+// CHECK-LABEL: define <2 x i64> @test_vcombine_p64(<1 x i64> %low, <1 x i64> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <1 x i64> %low, <1 x i64> %high, <2 x i32> <i32 0, i32 1>
+// CHECK: ret <2 x i64> [[SHUFFLE_I]]
poly64x2_t test_vcombine_p64(poly64x1_t low, poly64x1_t high) {
- // CHECK-LABEL: test_vcombine_p64:
return vcombine_p64(low, high);
- // CHECK: ins v0.d[1], v1.d[0]
}
+// CHECK-LABEL: define <8 x half> @test_vcombine_f16(<4 x half> %low, <4 x half> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <4 x half> %low, <4 x half> %high, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+// CHECK: ret <8 x half> [[SHUFFLE_I]]
float16x8_t test_vcombine_f16(float16x4_t low, float16x4_t high) {
- // CHECK-LABEL: test_vcombine_f16:
return vcombine_f16(low, high);
- // CHECK: ins v0.d[1], v1.d[0]
}
+// CHECK-LABEL: define <4 x float> @test_vcombine_f32(<2 x float> %low, <2 x float> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x float> %low, <2 x float> %high, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+// CHECK: ret <4 x float> [[SHUFFLE_I]]
float32x4_t test_vcombine_f32(float32x2_t low, float32x2_t high) {
- // CHECK-LABEL: test_vcombine_f32:
return vcombine_f32(low, high);
- // CHECK: ins v0.d[1], v1.d[0]
}
+// CHECK-LABEL: define <16 x i8> @test_vcombine_p8(<8 x i8> %low, <8 x i8> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x i8> %low, <8 x i8> %high, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: ret <16 x i8> [[SHUFFLE_I]]
poly8x16_t test_vcombine_p8(poly8x8_t low, poly8x8_t high) {
- // CHECK-LABEL: test_vcombine_p8:
return vcombine_p8(low, high);
- // CHECK: ins v0.d[1], v1.d[0]
}
+// CHECK-LABEL: define <8 x i16> @test_vcombine_p16(<4 x i16> %low, <4 x i16> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <4 x i16> %low, <4 x i16> %high, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+// CHECK: ret <8 x i16> [[SHUFFLE_I]]
poly16x8_t test_vcombine_p16(poly16x4_t low, poly16x4_t high) {
- // CHECK-LABEL: test_vcombine_p16:
return vcombine_p16(low, high);
- // CHECK: ins v0.d[1], v1.d[0]
}
+// CHECK-LABEL: define <2 x double> @test_vcombine_f64(<1 x double> %low, <1 x double> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <1 x double> %low, <1 x double> %high, <2 x i32> <i32 0, i32 1>
+// CHECK: ret <2 x double> [[SHUFFLE_I]]
float64x2_t test_vcombine_f64(float64x1_t low, float64x1_t high) {
- // CHECK-LABEL: test_vcombine_f64:
return vcombine_f64(low, high);
- // CHECK: ins v0.d[1], v1.d[0]
}
Modified: cfe/trunk/test/CodeGen/aarch64-neon-vget-hilo.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/aarch64-neon-vget-hilo.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/aarch64-neon-vget-hilo.c (original)
+++ cfe/trunk/test/CodeGen/aarch64-neon-vget-hilo.c Wed Mar 9 12:54:42 2016
@@ -1,176 +1,203 @@
-// REQUIRES: aarch64-registered-target
// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon \
-// RUN: -ffp-contract=fast -S -O3 -o - %s | FileCheck %s --check-prefix CHECK-COMMON --check-prefix CHECK-ARM64
-
+// RUN: -fallow-half-arguments-and-returns -emit-llvm -o - %s \
+// RUN: | opt -S -mem2reg | FileCheck %s
// Test new aarch64 intrinsics and types
#include <arm_neon.h>
+// CHECK-LABEL: define <8 x i8> @test_vget_high_s8(<16 x i8> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: ret <8 x i8> [[SHUFFLE_I]]
int8x8_t test_vget_high_s8(int8x16_t a) {
- // CHECK-COMMON-LABEL: test_vget_high_s8:
return vget_high_s8(a);
- // CHECK-ARM64: ext v0.16b, v0.16b, v0.16b, #8
}
+// CHECK-LABEL: define <4 x i16> @test_vget_high_s16(<8 x i16> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+// CHECK: ret <4 x i16> [[SHUFFLE_I]]
int16x4_t test_vget_high_s16(int16x8_t a) {
- // CHECK-COMMON-LABEL: test_vget_high_s16:
return vget_high_s16(a);
- // CHECK-ARM64: ext v0.16b, v0.16b, v0.16b, #8
}
+// CHECK-LABEL: define <2 x i32> @test_vget_high_s32(<4 x i32> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <4 x i32> %a, <4 x i32> %a, <2 x i32> <i32 2, i32 3>
+// CHECK: ret <2 x i32> [[SHUFFLE_I]]
int32x2_t test_vget_high_s32(int32x4_t a) {
- // CHECK-COMMON-LABEL: test_vget_high_s32:
return vget_high_s32(a);
- // CHECK-ARM64: ext v0.16b, v0.16b, v0.16b, #8
}
+// CHECK-LABEL: define <1 x i64> @test_vget_high_s64(<2 x i64> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> <i32 1>
+// CHECK: ret <1 x i64> [[SHUFFLE_I]]
int64x1_t test_vget_high_s64(int64x2_t a) {
- // CHECK-COMMON-LABEL: test_vget_high_s64:
return vget_high_s64(a);
- // CHECK-ARM64: ext v0.16b, v0.16b, v0.16b, #8
}
+// CHECK-LABEL: define <8 x i8> @test_vget_high_u8(<16 x i8> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: ret <8 x i8> [[SHUFFLE_I]]
uint8x8_t test_vget_high_u8(uint8x16_t a) {
- // CHECK-COMMON-LABEL: test_vget_high_u8:
return vget_high_u8(a);
- // CHECK-ARM64: ext v0.16b, v0.16b, v0.16b, #8
}
+// CHECK-LABEL: define <4 x i16> @test_vget_high_u16(<8 x i16> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+// CHECK: ret <4 x i16> [[SHUFFLE_I]]
uint16x4_t test_vget_high_u16(uint16x8_t a) {
- // CHECK-COMMON-LABEL: test_vget_high_u16:
return vget_high_u16(a);
- // CHECK-ARM64: ext v0.16b, v0.16b, v0.16b, #8
}
+// CHECK-LABEL: define <2 x i32> @test_vget_high_u32(<4 x i32> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <4 x i32> %a, <4 x i32> %a, <2 x i32> <i32 2, i32 3>
+// CHECK: ret <2 x i32> [[SHUFFLE_I]]
uint32x2_t test_vget_high_u32(uint32x4_t a) {
- // CHECK-COMMON-LABEL: test_vget_high_u32:
return vget_high_u32(a);
- // CHECK-ARM64: ext v0.16b, v0.16b, v0.16b, #8
}
+// CHECK-LABEL: define <1 x i64> @test_vget_high_u64(<2 x i64> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> <i32 1>
+// CHECK: ret <1 x i64> [[SHUFFLE_I]]
uint64x1_t test_vget_high_u64(uint64x2_t a) {
- // CHECK-COMMON-LABEL: test_vget_high_u64:
return vget_high_u64(a);
- // CHECK-ARM64: ext v0.16b, v0.16b, v0.16b, #8
}
+// CHECK-LABEL: define <1 x i64> @test_vget_high_p64(<2 x i64> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> <i32 1>
+// CHECK: ret <1 x i64> [[SHUFFLE_I]]
poly64x1_t test_vget_high_p64(poly64x2_t a) {
- // CHECK-COMMON-LABEL: test_vget_high_p64:
return vget_high_p64(a);
- // CHECK-ARM64: ext v0.16b, v0.16b, v0.16b, #8
}
+// CHECK-LABEL: define <4 x half> @test_vget_high_f16(<8 x half> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x half> %a, <8 x half> %a, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+// CHECK: ret <4 x half> [[SHUFFLE_I]]
float16x4_t test_vget_high_f16(float16x8_t a) {
- // CHECK-COMMON-LABEL: test_vget_high_f16:
return vget_high_f16(a);
- // CHECK-ARM64: ext v0.16b, v0.16b, v0.16b, #8
}
+// CHECK-LABEL: define <2 x float> @test_vget_high_f32(<4 x float> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <4 x float> %a, <4 x float> %a, <2 x i32> <i32 2, i32 3>
+// CHECK: ret <2 x float> [[SHUFFLE_I]]
float32x2_t test_vget_high_f32(float32x4_t a) {
- // CHECK-COMMON-LABEL: test_vget_high_f32:
return vget_high_f32(a);
- // CHECK-ARM64: ext v0.16b, v0.16b, v0.16b, #8
}
+// CHECK-LABEL: define <8 x i8> @test_vget_high_p8(<16 x i8> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: ret <8 x i8> [[SHUFFLE_I]]
poly8x8_t test_vget_high_p8(poly8x16_t a) {
- // CHECK-COMMON-LABEL: test_vget_high_p8:
return vget_high_p8(a);
- // CHECK-ARM64: ext v0.16b, v0.16b, v0.16b, #8
}
+// CHECK-LABEL: define <4 x i16> @test_vget_high_p16(<8 x i16> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+// CHECK: ret <4 x i16> [[SHUFFLE_I]]
poly16x4_t test_vget_high_p16(poly16x8_t a) {
- // CHECK-COMMON-LABEL: test_vget_high_p16
return vget_high_p16(a);
- // CHECK-ARM64: ext v0.16b, v0.16b, v0.16b, #8
}
+// CHECK-LABEL: define <1 x double> @test_vget_high_f64(<2 x double> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x double> %a, <2 x double> %a, <1 x i32> <i32 1>
+// CHECK: ret <1 x double> [[SHUFFLE_I]]
float64x1_t test_vget_high_f64(float64x2_t a) {
- // CHECK-COMMON-LABEL: test_vget_high_f64
return vget_high_f64(a);
- // CHECK-ARM64: ext v0.16b, v0.16b, v0.16b, #8
}
+// CHECK-LABEL: define <8 x i8> @test_vget_low_s8(<16 x i8> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+// CHECK: ret <8 x i8> [[SHUFFLE_I]]
int8x8_t test_vget_low_s8(int8x16_t a) {
- // CHECK-COMMON-LABEL: test_vget_low_s8:
return vget_low_s8(a);
- // CHECK-COMMON-NEXT: ret
}
+// CHECK-LABEL: define <4 x i16> @test_vget_low_s16(<8 x i16> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+// CHECK: ret <4 x i16> [[SHUFFLE_I]]
int16x4_t test_vget_low_s16(int16x8_t a) {
- // CHECK-COMMON-LABEL: test_vget_low_s16:
return vget_low_s16(a);
- // CHECK-COMMON-NEXT: ret
}
+// CHECK-LABEL: define <2 x i32> @test_vget_low_s32(<4 x i32> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <4 x i32> %a, <4 x i32> %a, <2 x i32> <i32 0, i32 1>
+// CHECK: ret <2 x i32> [[SHUFFLE_I]]
int32x2_t test_vget_low_s32(int32x4_t a) {
- // CHECK-COMMON-LABEL: test_vget_low_s32:
return vget_low_s32(a);
- // CHECK-COMMON-NEXT: ret
}
+// CHECK-LABEL: define <1 x i64> @test_vget_low_s64(<2 x i64> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> zeroinitializer
+// CHECK: ret <1 x i64> [[SHUFFLE_I]]
int64x1_t test_vget_low_s64(int64x2_t a) {
- // CHECK-COMMON-LABEL: test_vget_low_s64:
return vget_low_s64(a);
- // CHECK-COMMON-NEXT: ret
}
+// CHECK-LABEL: define <8 x i8> @test_vget_low_u8(<16 x i8> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+// CHECK: ret <8 x i8> [[SHUFFLE_I]]
uint8x8_t test_vget_low_u8(uint8x16_t a) {
- // CHECK-COMMON-LABEL: test_vget_low_u8:
return vget_low_u8(a);
- // CHECK-COMMON-NEXT: ret
}
+// CHECK-LABEL: define <4 x i16> @test_vget_low_u16(<8 x i16> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+// CHECK: ret <4 x i16> [[SHUFFLE_I]]
uint16x4_t test_vget_low_u16(uint16x8_t a) {
- // CHECK-COMMON-LABEL: test_vget_low_u16:
return vget_low_u16(a);
- // CHECK-COMMON-NEXT: ret
}
+// CHECK-LABEL: define <2 x i32> @test_vget_low_u32(<4 x i32> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <4 x i32> %a, <4 x i32> %a, <2 x i32> <i32 0, i32 1>
+// CHECK: ret <2 x i32> [[SHUFFLE_I]]
uint32x2_t test_vget_low_u32(uint32x4_t a) {
- // CHECK-COMMON-LABEL: test_vget_low_u32:
return vget_low_u32(a);
- // CHECK-COMMON-NEXT: ret
}
+// CHECK-LABEL: define <1 x i64> @test_vget_low_u64(<2 x i64> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> zeroinitializer
+// CHECK: ret <1 x i64> [[SHUFFLE_I]]
uint64x1_t test_vget_low_u64(uint64x2_t a) {
- // CHECK-COMMON-LABEL: test_vget_low_u64:
return vget_low_u64(a);
- // CHECK-COMMON-NEXT: ret
}
+// CHECK-LABEL: define <1 x i64> @test_vget_low_p64(<2 x i64> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> zeroinitializer
+// CHECK: ret <1 x i64> [[SHUFFLE_I]]
poly64x1_t test_vget_low_p64(poly64x2_t a) {
- // CHECK-COMMON-LABEL: test_vget_low_p64:
return vget_low_p64(a);
- // CHECK-COMMON-NEXT: ret
}
+// CHECK-LABEL: define <4 x half> @test_vget_low_f16(<8 x half> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x half> %a, <8 x half> %a, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+// CHECK: ret <4 x half> [[SHUFFLE_I]]
float16x4_t test_vget_low_f16(float16x8_t a) {
- // CHECK-COMMON-LABEL: test_vget_low_f16:
return vget_low_f16(a);
- // CHECK-COMMON-NEXT: ret
}
+// CHECK-LABEL: define <2 x float> @test_vget_low_f32(<4 x float> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <4 x float> %a, <4 x float> %a, <2 x i32> <i32 0, i32 1>
+// CHECK: ret <2 x float> [[SHUFFLE_I]]
float32x2_t test_vget_low_f32(float32x4_t a) {
- // CHECK-COMMON-LABEL: test_vget_low_f32:
return vget_low_f32(a);
- // CHECK-COMMON-NEXT: ret
}
+// CHECK-LABEL: define <8 x i8> @test_vget_low_p8(<16 x i8> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+// CHECK: ret <8 x i8> [[SHUFFLE_I]]
poly8x8_t test_vget_low_p8(poly8x16_t a) {
- // CHECK-COMMON-LABEL: test_vget_low_p8:
return vget_low_p8(a);
- // CHECK-COMMON-NEXT: ret
}
+// CHECK-LABEL: define <4 x i16> @test_vget_low_p16(<8 x i16> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+// CHECK: ret <4 x i16> [[SHUFFLE_I]]
poly16x4_t test_vget_low_p16(poly16x8_t a) {
- // CHECK-COMMON-LABEL: test_vget_low_p16:
return vget_low_p16(a);
- // CHECK-COMMON-NEXT: ret
}
+// CHECK-LABEL: define <1 x double> @test_vget_low_f64(<2 x double> %a) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x double> %a, <2 x double> %a, <1 x i32> zeroinitializer
+// CHECK: ret <1 x double> [[SHUFFLE_I]]
float64x1_t test_vget_low_f64(float64x2_t a) {
- // CHECK-COMMON-LABEL: test_vget_low_f64:
return vget_low_f64(a);
- // CHECK-COMMON-NEXT: ret
}
Modified: cfe/trunk/test/CodeGen/aarch64-neon-vget.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/aarch64-neon-vget.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/aarch64-neon-vget.c (original)
+++ cfe/trunk/test/CodeGen/aarch64-neon-vget.c Wed Mar 9 12:54:42 2016
@@ -1,348 +1,458 @@
-// REQUIRES: aarch64-registered-target
// RUN: %clang_cc1 -triple arm64-apple-darwin -target-feature +neon \
-// RUN: -ffp-contract=fast -S -O3 -o - %s | FileCheck %s
+// RUN: -fallow-half-arguments-and-returns -emit-llvm -o - %s \
+// RUN: | opt -S -mem2reg | FileCheck %s
#include <arm_neon.h>
+// CHECK-LABEL: define i8 @test_vget_lane_u8(<8 x i8> %a) #0 {
+// CHECK: [[VGET_LANE:%.*]] = extractelement <8 x i8> %a, i32 7
+// CHECK: ret i8 [[VGET_LANE]]
uint8_t test_vget_lane_u8(uint8x8_t a) {
- // CHECK-LABEL: test_vget_lane_u8:
- // CHECK-NEXT: umov.b w0, v0[7]
- // CHECK-NEXT: ret
return vget_lane_u8(a, 7);
}
+// CHECK-LABEL: define i16 @test_vget_lane_u16(<4 x i16> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <4 x i16> [[TMP1]], i32 3
+// CHECK: ret i16 [[VGET_LANE]]
uint16_t test_vget_lane_u16(uint16x4_t a) {
- // CHECK-LABEL: test_vget_lane_u16:
- // CHECK-NEXT: umov.h w0, v0[3]
- // CHECK-NEXT: ret
return vget_lane_u16(a, 3);
}
+// CHECK-LABEL: define i32 @test_vget_lane_u32(<2 x i32> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <2 x i32> [[TMP1]], i32 1
+// CHECK: ret i32 [[VGET_LANE]]
uint32_t test_vget_lane_u32(uint32x2_t a) {
- // CHECK-LABEL: test_vget_lane_u32:
- // CHECK-NEXT: mov.s w0, v0[1]
- // CHECK-NEXT: ret
return vget_lane_u32(a, 1);
}
+// CHECK-LABEL: define i8 @test_vget_lane_s8(<8 x i8> %a) #0 {
+// CHECK: [[VGET_LANE:%.*]] = extractelement <8 x i8> %a, i32 7
+// CHECK: ret i8 [[VGET_LANE]]
int8_t test_vget_lane_s8(int8x8_t a) {
- // CHECK-LABEL: test_vget_lane_s8:
- // CHECK-NEXT: umov.b w0, v0[7]
- // CHECK-NEXT: ret
return vget_lane_s8(a, 7);
}
+// CHECK-LABEL: define i16 @test_vget_lane_s16(<4 x i16> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <4 x i16> [[TMP1]], i32 3
+// CHECK: ret i16 [[VGET_LANE]]
int16_t test_vget_lane_s16(int16x4_t a) {
- // CHECK-LABEL: test_vget_lane_s16:
- // CHECK-NEXT: umov.h w0, v0[3]
- // CHECK-NEXT: ret
return vget_lane_s16(a, 3);
}
+// CHECK-LABEL: define i32 @test_vget_lane_s32(<2 x i32> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <2 x i32> [[TMP1]], i32 1
+// CHECK: ret i32 [[VGET_LANE]]
int32_t test_vget_lane_s32(int32x2_t a) {
- // CHECK-LABEL: test_vget_lane_s32:
- // CHECK-NEXT: mov.s w0, v0[1]
- // CHECK-NEXT: ret
return vget_lane_s32(a, 1);
}
+// CHECK-LABEL: define i8 @test_vget_lane_p8(<8 x i8> %a) #0 {
+// CHECK: [[VGET_LANE:%.*]] = extractelement <8 x i8> %a, i32 7
+// CHECK: ret i8 [[VGET_LANE]]
poly8_t test_vget_lane_p8(poly8x8_t a) {
- // CHECK-LABEL: test_vget_lane_p8:
- // CHECK-NEXT: umov.b w0, v0[7]
- // CHECK-NEXT: ret
return vget_lane_p8(a, 7);
}
+// CHECK-LABEL: define i16 @test_vget_lane_p16(<4 x i16> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <4 x i16> [[TMP1]], i32 3
+// CHECK: ret i16 [[VGET_LANE]]
poly16_t test_vget_lane_p16(poly16x4_t a) {
- // CHECK-LABEL: test_vget_lane_p16:
- // CHECK-NEXT: umov.h w0, v0[3]
- // CHECK-NEXT: ret
return vget_lane_p16(a, 3);
}
+// CHECK-LABEL: define float @test_vget_lane_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <2 x float> [[TMP1]], i32 1
+// CHECK: ret float [[VGET_LANE]]
float32_t test_vget_lane_f32(float32x2_t a) {
- // CHECK-LABEL: test_vget_lane_f32:
- // CHECK-NEXT: mov s0, v0[1]
- // CHECK-NEXT: ret
return vget_lane_f32(a, 1);
}
+// CHECK-LABEL: define float @test_vget_lane_f16(<4 x half> %a) #0 {
+// CHECK: [[__REINT_242:%.*]] = alloca <4 x half>, align 8
+// CHECK: [[__REINT1_242:%.*]] = alloca i16, align 2
+// CHECK: store <4 x half> %a, <4 x half>* [[__REINT_242]], align 8
+// CHECK: [[TMP0:%.*]] = bitcast <4 x half>* [[__REINT_242]] to <4 x i16>*
+// CHECK: [[TMP1:%.*]] = load <4 x i16>, <4 x i16>* [[TMP0]], align 8
+// CHECK: [[TMP2:%.*]] = bitcast <4 x i16> [[TMP1]] to <8 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP2]] to <4 x i16>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <4 x i16> [[TMP3]], i32 1
+// CHECK: store i16 [[VGET_LANE]], i16* [[__REINT1_242]], align 2
+// CHECK: [[TMP4:%.*]] = bitcast i16* [[__REINT1_242]] to half*
+// CHECK: [[TMP5:%.*]] = load half, half* [[TMP4]], align 2
+// CHECK: [[CONV:%.*]] = fpext half [[TMP5]] to float
+// CHECK: ret float [[CONV]]
float32_t test_vget_lane_f16(float16x4_t a) {
- // CHECK-LABEL: test_vget_lane_f16:
- // CHECK-NEXT: umov.h w8, v0[1]
- // CHECK-NEXT: fmov s0, w8
- // CHECK-NEXT: fcvt s0, h0
- // CHECK-NEXT: ret
return vget_lane_f16(a, 1);
}
+// CHECK-LABEL: define i8 @test_vgetq_lane_u8(<16 x i8> %a) #0 {
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <16 x i8> %a, i32 15
+// CHECK: ret i8 [[VGETQ_LANE]]
uint8_t test_vgetq_lane_u8(uint8x16_t a) {
- // CHECK-LABEL: test_vgetq_lane_u8:
- // CHECK-NEXT: umov.b w0, v0[15]
- // CHECK-NEXT: ret
return vgetq_lane_u8(a, 15);
}
+// CHECK-LABEL: define i16 @test_vgetq_lane_u16(<8 x i16> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <8 x i16> [[TMP1]], i32 7
+// CHECK: ret i16 [[VGETQ_LANE]]
uint16_t test_vgetq_lane_u16(uint16x8_t a) {
- // CHECK-LABEL: test_vgetq_lane_u16:
- // CHECK-NEXT: umov.h w0, v0[7]
- // CHECK-NEXT: ret
return vgetq_lane_u16(a, 7);
}
+// CHECK-LABEL: define i32 @test_vgetq_lane_u32(<4 x i32> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <4 x i32> [[TMP1]], i32 3
+// CHECK: ret i32 [[VGETQ_LANE]]
uint32_t test_vgetq_lane_u32(uint32x4_t a) {
- // CHECK-LABEL: test_vgetq_lane_u32:
- // CHECK-NEXT: mov.s w0, v0[3]
- // CHECK-NEXT: ret
return vgetq_lane_u32(a, 3);
}
+// CHECK-LABEL: define i8 @test_vgetq_lane_s8(<16 x i8> %a) #0 {
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <16 x i8> %a, i32 15
+// CHECK: ret i8 [[VGETQ_LANE]]
int8_t test_vgetq_lane_s8(int8x16_t a) {
- // CHECK-LABEL: test_vgetq_lane_s8:
- // CHECK-NEXT: umov.b w0, v0[15]
- // CHECK-NEXT: ret
return vgetq_lane_s8(a, 15);
}
+// CHECK-LABEL: define i16 @test_vgetq_lane_s16(<8 x i16> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <8 x i16> [[TMP1]], i32 7
+// CHECK: ret i16 [[VGETQ_LANE]]
int16_t test_vgetq_lane_s16(int16x8_t a) {
- // CHECK-LABEL: test_vgetq_lane_s16:
- // CHECK-NEXT: umov.h w0, v0[7]
- // CHECK-NEXT: ret
return vgetq_lane_s16(a, 7);
}
+// CHECK-LABEL: define i32 @test_vgetq_lane_s32(<4 x i32> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <4 x i32> [[TMP1]], i32 3
+// CHECK: ret i32 [[VGETQ_LANE]]
int32_t test_vgetq_lane_s32(int32x4_t a) {
- // CHECK-LABEL: test_vgetq_lane_s32:
- // CHECK-NEXT: mov.s w0, v0[3]
- // CHECK-NEXT: ret
return vgetq_lane_s32(a, 3);
}
+// CHECK-LABEL: define i8 @test_vgetq_lane_p8(<16 x i8> %a) #0 {
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <16 x i8> %a, i32 15
+// CHECK: ret i8 [[VGETQ_LANE]]
poly8_t test_vgetq_lane_p8(poly8x16_t a) {
- // CHECK-LABEL: test_vgetq_lane_p8:
- // CHECK-NEXT: umov.b w0, v0[15]
- // CHECK-NEXT: ret
return vgetq_lane_p8(a, 15);
}
+// CHECK-LABEL: define i16 @test_vgetq_lane_p16(<8 x i16> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <8 x i16> [[TMP1]], i32 7
+// CHECK: ret i16 [[VGETQ_LANE]]
poly16_t test_vgetq_lane_p16(poly16x8_t a) {
- // CHECK-LABEL: test_vgetq_lane_p16:
- // CHECK-NEXT: umov.h w0, v0[7]
- // CHECK-NEXT: ret
return vgetq_lane_p16(a, 7);
}
+// CHECK-LABEL: define float @test_vgetq_lane_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <4 x float> [[TMP1]], i32 3
+// CHECK: ret float [[VGETQ_LANE]]
float32_t test_vgetq_lane_f32(float32x4_t a) {
- // CHECK-LABEL: test_vgetq_lane_f32:
- // CHECK-NEXT: mov s0, v0[3]
- // CHECK-NEXT: ret
return vgetq_lane_f32(a, 3);
}
+// CHECK-LABEL: define float @test_vgetq_lane_f16(<8 x half> %a) #0 {
+// CHECK: [[__REINT_244:%.*]] = alloca <8 x half>, align 16
+// CHECK: [[__REINT1_244:%.*]] = alloca i16, align 2
+// CHECK: store <8 x half> %a, <8 x half>* [[__REINT_244]], align 16
+// CHECK: [[TMP0:%.*]] = bitcast <8 x half>* [[__REINT_244]] to <8 x i16>*
+// CHECK: [[TMP1:%.*]] = load <8 x i16>, <8 x i16>* [[TMP0]], align 16
+// CHECK: [[TMP2:%.*]] = bitcast <8 x i16> [[TMP1]] to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP2]] to <8 x i16>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <8 x i16> [[TMP3]], i32 3
+// CHECK: store i16 [[VGETQ_LANE]], i16* [[__REINT1_244]], align 2
+// CHECK: [[TMP4:%.*]] = bitcast i16* [[__REINT1_244]] to half*
+// CHECK: [[TMP5:%.*]] = load half, half* [[TMP4]], align 2
+// CHECK: [[CONV:%.*]] = fpext half [[TMP5]] to float
+// CHECK: ret float [[CONV]]
float32_t test_vgetq_lane_f16(float16x8_t a) {
- // CHECK-LABEL: test_vgetq_lane_f16:
- // CHECK-NEXT: umov.h w8, v0[3]
- // CHECK-NEXT: fmov s0, w8
- // CHECK-NEXT: fcvt s0, h0
- // CHECK-NEXT: ret
return vgetq_lane_f16(a, 3);
}
+// CHECK-LABEL: define i64 @test_vget_lane_s64(<1 x i64> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <1 x i64> [[TMP1]], i32 0
+// CHECK: ret i64 [[VGET_LANE]]
int64_t test_vget_lane_s64(int64x1_t a) {
- // CHECK-LABEL: test_vget_lane_s64:
- // CHECK-NEXT: fmov x0, d0
- // CHECK-NEXT: ret
return vget_lane_s64(a, 0);
}
+// CHECK-LABEL: define i64 @test_vget_lane_u64(<1 x i64> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <1 x i64> [[TMP1]], i32 0
+// CHECK: ret i64 [[VGET_LANE]]
uint64_t test_vget_lane_u64(uint64x1_t a) {
- // CHECK-LABEL: test_vget_lane_u64:
- // CHECK-NEXT: fmov x0, d0
- // CHECK-NEXT: ret
return vget_lane_u64(a, 0);
}
+// CHECK-LABEL: define i64 @test_vgetq_lane_s64(<2 x i64> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <2 x i64> [[TMP1]], i32 1
+// CHECK: ret i64 [[VGETQ_LANE]]
int64_t test_vgetq_lane_s64(int64x2_t a) {
- // CHECK-LABEL: test_vgetq_lane_s64:
- // CHECK-NEXT: mov.d x0, v0[1]
- // CHECK-NEXT: ret
return vgetq_lane_s64(a, 1);
}
+// CHECK-LABEL: define i64 @test_vgetq_lane_u64(<2 x i64> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <2 x i64> [[TMP1]], i32 1
+// CHECK: ret i64 [[VGETQ_LANE]]
uint64_t test_vgetq_lane_u64(uint64x2_t a) {
- // CHECK-LABEL: test_vgetq_lane_u64:
- // CHECK-NEXT: mov.d x0, v0[1]
- // CHECK-NEXT: ret
return vgetq_lane_u64(a, 1);
}
+// CHECK-LABEL: define <8 x i8> @test_vset_lane_u8(i8 %a, <8 x i8> %b) #0 {
+// CHECK: [[VSET_LANE:%.*]] = insertelement <8 x i8> %b, i8 %a, i32 7
+// CHECK: ret <8 x i8> [[VSET_LANE]]
uint8x8_t test_vset_lane_u8(uint8_t a, uint8x8_t b) {
- // CHECK-LABEL: test_vset_lane_u8:
- // CHECK-NEXT: ins.b v0[7], w0
- // CHECK-NEXT: ret
return vset_lane_u8(a, b, 7);
}
+// CHECK-LABEL: define <4 x i16> @test_vset_lane_u16(i16 %a, <4 x i16> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <4 x i16> [[TMP1]], i16 %a, i32 3
+// CHECK: ret <4 x i16> [[VSET_LANE]]
uint16x4_t test_vset_lane_u16(uint16_t a, uint16x4_t b) {
- // CHECK-LABEL: test_vset_lane_u16:
- // CHECK-NEXT: ins.h v0[3], w0
- // CHECK-NEXT: ret
return vset_lane_u16(a, b, 3);
}
+// CHECK-LABEL: define <2 x i32> @test_vset_lane_u32(i32 %a, <2 x i32> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <2 x i32> [[TMP1]], i32 %a, i32 1
+// CHECK: ret <2 x i32> [[VSET_LANE]]
uint32x2_t test_vset_lane_u32(uint32_t a, uint32x2_t b) {
- // CHECK-LABEL: test_vset_lane_u32:
- // CHECK-NEXT: ins.s v0[1], w0
- // CHECK-NEXT: ret
return vset_lane_u32(a, b, 1);
}
+// CHECK-LABEL: define <8 x i8> @test_vset_lane_s8(i8 %a, <8 x i8> %b) #0 {
+// CHECK: [[VSET_LANE:%.*]] = insertelement <8 x i8> %b, i8 %a, i32 7
+// CHECK: ret <8 x i8> [[VSET_LANE]]
int8x8_t test_vset_lane_s8(int8_t a, int8x8_t b) {
- // CHECK-LABEL: test_vset_lane_s8:
- // CHECK-NEXT: ins.b v0[7], w0
- // CHECK-NEXT: ret
return vset_lane_s8(a, b, 7);
}
+// CHECK-LABEL: define <4 x i16> @test_vset_lane_s16(i16 %a, <4 x i16> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <4 x i16> [[TMP1]], i16 %a, i32 3
+// CHECK: ret <4 x i16> [[VSET_LANE]]
int16x4_t test_vset_lane_s16(int16_t a, int16x4_t b) {
- // CHECK-LABEL: test_vset_lane_s16:
- // CHECK-NEXT: ins.h v0[3], w0
- // CHECK-NEXT: ret
return vset_lane_s16(a, b, 3);
}
+// CHECK-LABEL: define <2 x i32> @test_vset_lane_s32(i32 %a, <2 x i32> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i32> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x i32>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <2 x i32> [[TMP1]], i32 %a, i32 1
+// CHECK: ret <2 x i32> [[VSET_LANE]]
int32x2_t test_vset_lane_s32(int32_t a, int32x2_t b) {
- // CHECK-LABEL: test_vset_lane_s32:
- // CHECK-NEXT: ins.s v0[1], w0
- // CHECK-NEXT: ret
return vset_lane_s32(a, b, 1);
}
+// CHECK-LABEL: define <8 x i8> @test_vset_lane_p8(i8 %a, <8 x i8> %b) #0 {
+// CHECK: [[VSET_LANE:%.*]] = insertelement <8 x i8> %b, i8 %a, i32 7
+// CHECK: ret <8 x i8> [[VSET_LANE]]
poly8x8_t test_vset_lane_p8(poly8_t a, poly8x8_t b) {
- // CHECK-LABEL: test_vset_lane_p8:
- // CHECK-NEXT: ins.b v0[7], w0
- // CHECK-NEXT: ret
return vset_lane_p8(a, b, 7);
}
+// CHECK-LABEL: define <4 x i16> @test_vset_lane_p16(i16 %a, <4 x i16> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i16> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <4 x i16>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <4 x i16> [[TMP1]], i16 %a, i32 3
+// CHECK: ret <4 x i16> [[VSET_LANE]]
poly16x4_t test_vset_lane_p16(poly16_t a, poly16x4_t b) {
- // CHECK-LABEL: test_vset_lane_p16:
- // CHECK-NEXT: ins.h v0[3], w0
- // CHECK-NEXT: ret
return vset_lane_p16(a, b, 3);
}
+// CHECK-LABEL: define <2 x float> @test_vset_lane_f32(float %a, <2 x float> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <2 x float> [[TMP1]], float %a, i32 1
+// CHECK: ret <2 x float> [[VSET_LANE]]
float32x2_t test_vset_lane_f32(float32_t a, float32x2_t b) {
- // CHECK-LABEL: test_vset_lane_f32:
- // CHECK-NEXT: ins.s v1[1], v0[0]
- // CHECK-NEXT: mov.16b v0, v1
- // CHECK-NEXT: ret
return vset_lane_f32(a, b, 1);
}
+// CHECK-LABEL: define <4 x half> @test_vset_lane_f16(half* %a, <4 x half> %b) #0 {
+// CHECK: [[__REINT_246:%.*]] = alloca half, align 2
+// CHECK: [[__REINT1_246:%.*]] = alloca <4 x half>, align 8
+// CHECK: [[__REINT2_246:%.*]] = alloca <4 x i16>, align 8
+// CHECK: [[TMP0:%.*]] = load half, half* %a, align 2
+// CHECK: store half [[TMP0]], half* [[__REINT_246]], align 2
+// CHECK: store <4 x half> %b, <4 x half>* [[__REINT1_246]], align 8
+// CHECK: [[TMP1:%.*]] = bitcast half* [[__REINT_246]] to i16*
+// CHECK: [[TMP2:%.*]] = load i16, i16* [[TMP1]], align 2
+// CHECK: [[TMP3:%.*]] = bitcast <4 x half>* [[__REINT1_246]] to <4 x i16>*
+// CHECK: [[TMP4:%.*]] = load <4 x i16>, <4 x i16>* [[TMP3]], align 8
+// CHECK: [[TMP5:%.*]] = bitcast <4 x i16> [[TMP4]] to <8 x i8>
+// CHECK: [[TMP6:%.*]] = bitcast <8 x i8> [[TMP5]] to <4 x i16>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <4 x i16> [[TMP6]], i16 [[TMP2]], i32 3
+// CHECK: store <4 x i16> [[VSET_LANE]], <4 x i16>* [[__REINT2_246]], align 8
+// CHECK: [[TMP7:%.*]] = bitcast <4 x i16>* [[__REINT2_246]] to <4 x half>*
+// CHECK: [[TMP8:%.*]] = load <4 x half>, <4 x half>* [[TMP7]], align 8
+// CHECK: ret <4 x half> [[TMP8]]
float16x4_t test_vset_lane_f16(float16_t *a, float16x4_t b) {
- // CHECK-LABEL: test_vset_lane_f16:
- // CHECK-NEXT: ld1.h { v0 }[3], [x0]
- // CHECK-NEXT: ret
return vset_lane_f16(*a, b, 3);
}
+// CHECK-LABEL: define <16 x i8> @test_vsetq_lane_u8(i8 %a, <16 x i8> %b) #0 {
+// CHECK: [[VSET_LANE:%.*]] = insertelement <16 x i8> %b, i8 %a, i32 15
+// CHECK: ret <16 x i8> [[VSET_LANE]]
uint8x16_t test_vsetq_lane_u8(uint8_t a, uint8x16_t b) {
- // CHECK-LABEL: test_vsetq_lane_u8:
- // CHECK-NEXT: ins.b v0[15], w0
- // CHECK-NEXT: ret
return vsetq_lane_u8(a, b, 15);
}
+// CHECK-LABEL: define <8 x i16> @test_vsetq_lane_u16(i16 %a, <8 x i16> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <8 x i16> [[TMP1]], i16 %a, i32 7
+// CHECK: ret <8 x i16> [[VSET_LANE]]
uint16x8_t test_vsetq_lane_u16(uint16_t a, uint16x8_t b) {
- // CHECK-LABEL: test_vsetq_lane_u16:
- // CHECK-NEXT: ins.h v0[7], w0
- // CHECK-NEXT: ret
return vsetq_lane_u16(a, b, 7);
}
+// CHECK-LABEL: define <4 x i32> @test_vsetq_lane_u32(i32 %a, <4 x i32> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <4 x i32> [[TMP1]], i32 %a, i32 3
+// CHECK: ret <4 x i32> [[VSET_LANE]]
uint32x4_t test_vsetq_lane_u32(uint32_t a, uint32x4_t b) {
- // CHECK-LABEL: test_vsetq_lane_u32:
- // CHECK-NEXT: ins.s v0[3], w0
- // CHECK-NEXT: ret
return vsetq_lane_u32(a, b, 3);
}
+// CHECK-LABEL: define <16 x i8> @test_vsetq_lane_s8(i8 %a, <16 x i8> %b) #0 {
+// CHECK: [[VSET_LANE:%.*]] = insertelement <16 x i8> %b, i8 %a, i32 15
+// CHECK: ret <16 x i8> [[VSET_LANE]]
int8x16_t test_vsetq_lane_s8(int8_t a, int8x16_t b) {
- // CHECK-LABEL: test_vsetq_lane_s8:
- // CHECK-NEXT: ins.b v0[15], w0
- // CHECK-NEXT: ret
return vsetq_lane_s8(a, b, 15);
}
+// CHECK-LABEL: define <8 x i16> @test_vsetq_lane_s16(i16 %a, <8 x i16> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <8 x i16> [[TMP1]], i16 %a, i32 7
+// CHECK: ret <8 x i16> [[VSET_LANE]]
int16x8_t test_vsetq_lane_s16(int16_t a, int16x8_t b) {
- // CHECK-LABEL: test_vsetq_lane_s16:
- // CHECK-NEXT: ins.h v0[7], w0
- // CHECK-NEXT: ret
return vsetq_lane_s16(a, b, 7);
}
+// CHECK-LABEL: define <4 x i32> @test_vsetq_lane_s32(i32 %a, <4 x i32> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <4 x i32> [[TMP1]], i32 %a, i32 3
+// CHECK: ret <4 x i32> [[VSET_LANE]]
int32x4_t test_vsetq_lane_s32(int32_t a, int32x4_t b) {
- // CHECK-LABEL: test_vsetq_lane_s32:
- // CHECK-NEXT: ins.s v0[3], w0
- // CHECK-NEXT: ret
return vsetq_lane_s32(a, b, 3);
}
+// CHECK-LABEL: define <16 x i8> @test_vsetq_lane_p8(i8 %a, <16 x i8> %b) #0 {
+// CHECK: [[VSET_LANE:%.*]] = insertelement <16 x i8> %b, i8 %a, i32 15
+// CHECK: ret <16 x i8> [[VSET_LANE]]
poly8x16_t test_vsetq_lane_p8(poly8_t a, poly8x16_t b) {
- // CHECK-LABEL: test_vsetq_lane_p8:
- // CHECK-NEXT: ins.b v0[15], w0
- // CHECK-NEXT: ret
return vsetq_lane_p8(a, b, 15);
}
+// CHECK-LABEL: define <8 x i16> @test_vsetq_lane_p16(i16 %a, <8 x i16> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <8 x i16> [[TMP1]], i16 %a, i32 7
+// CHECK: ret <8 x i16> [[VSET_LANE]]
poly16x8_t test_vsetq_lane_p16(poly16_t a, poly16x8_t b) {
- // CHECK-LABEL: test_vsetq_lane_p16:
- // CHECK-NEXT: ins.h v0[7], w0
- // CHECK-NEXT: ret
return vsetq_lane_p16(a, b, 7);
}
+// CHECK-LABEL: define <4 x float> @test_vsetq_lane_f32(float %a, <4 x float> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <4 x float> [[TMP1]], float %a, i32 3
+// CHECK: ret <4 x float> [[VSET_LANE]]
float32x4_t test_vsetq_lane_f32(float32_t a, float32x4_t b) {
- // CHECK-LABEL: test_vsetq_lane_f32:
- // CHECK-NEXT: ins.s v1[3], v0[0]
- // CHECK-NEXT: mov.16b v0, v1
- // CHECK-NEXT: ret
return vsetq_lane_f32(a, b, 3);
}
+// CHECK-LABEL: define <8 x half> @test_vsetq_lane_f16(half* %a, <8 x half> %b) #0 {
+// CHECK: [[__REINT_248:%.*]] = alloca half, align 2
+// CHECK: [[__REINT1_248:%.*]] = alloca <8 x half>, align 16
+// CHECK: [[__REINT2_248:%.*]] = alloca <8 x i16>, align 16
+// CHECK: [[TMP0:%.*]] = load half, half* %a, align 2
+// CHECK: store half [[TMP0]], half* [[__REINT_248]], align 2
+// CHECK: store <8 x half> %b, <8 x half>* [[__REINT1_248]], align 16
+// CHECK: [[TMP1:%.*]] = bitcast half* [[__REINT_248]] to i16*
+// CHECK: [[TMP2:%.*]] = load i16, i16* [[TMP1]], align 2
+// CHECK: [[TMP3:%.*]] = bitcast <8 x half>* [[__REINT1_248]] to <8 x i16>*
+// CHECK: [[TMP4:%.*]] = load <8 x i16>, <8 x i16>* [[TMP3]], align 16
+// CHECK: [[TMP5:%.*]] = bitcast <8 x i16> [[TMP4]] to <16 x i8>
+// CHECK: [[TMP6:%.*]] = bitcast <16 x i8> [[TMP5]] to <8 x i16>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <8 x i16> [[TMP6]], i16 [[TMP2]], i32 7
+// CHECK: store <8 x i16> [[VSET_LANE]], <8 x i16>* [[__REINT2_248]], align 16
+// CHECK: [[TMP7:%.*]] = bitcast <8 x i16>* [[__REINT2_248]] to <8 x half>*
+// CHECK: [[TMP8:%.*]] = load <8 x half>, <8 x half>* [[TMP7]], align 16
+// CHECK: ret <8 x half> [[TMP8]]
float16x8_t test_vsetq_lane_f16(float16_t *a, float16x8_t b) {
- // CHECK-LABEL: test_vsetq_lane_f16:
- // CHECK-NEXT: ld1.h { v0 }[7], [x0]
- // CHECK-NEXT: ret
return vsetq_lane_f16(*a, b, 7);
}
+// CHECK-LABEL: define <1 x i64> @test_vset_lane_s64(i64 %a, <1 x i64> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <1 x i64> [[TMP1]], i64 %a, i32 0
+// CHECK: ret <1 x i64> [[VSET_LANE]]
int64x1_t test_vset_lane_s64(int64_t a, int64x1_t b) {
- // CHECK-LABEL: test_vset_lane_s64:
- // CHECK-NEXT: fmov d0, x0
- // CHECK-NEXT: ret
return vset_lane_s64(a, b, 0);
}
+// CHECK-LABEL: define <1 x i64> @test_vset_lane_u64(i64 %a, <1 x i64> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <1 x i64> [[TMP1]], i64 %a, i32 0
+// CHECK: ret <1 x i64> [[VSET_LANE]]
uint64x1_t test_vset_lane_u64(uint64_t a, uint64x1_t b) {
- // CHECK-LABEL: test_vset_lane_u64:
- // CHECK-NEXT: fmov d0, x0
- // CHECK-NEXT: ret
return vset_lane_u64(a, b, 0);
}
+// CHECK-LABEL: define <2 x i64> @test_vsetq_lane_s64(i64 %a, <2 x i64> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <2 x i64> [[TMP1]], i64 %a, i32 1
+// CHECK: ret <2 x i64> [[VSET_LANE]]
int64x2_t test_vsetq_lane_s64(int64_t a, int64x2_t b) {
- // CHECK-LABEL: test_vsetq_lane_s64:
- // CHECK-NEXT: ins.d v0[1], x0
- // CHECK-NEXT: ret
return vsetq_lane_s64(a, b, 1);
}
+// CHECK-LABEL: define <2 x i64> @test_vsetq_lane_u64(i64 %a, <2 x i64> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <2 x i64> [[TMP1]], i64 %a, i32 1
+// CHECK: ret <2 x i64> [[VSET_LANE]]
uint64x2_t test_vsetq_lane_u64(uint64_t a, uint64x2_t b) {
- // CHECK-LABEL: test_vsetq_lane_u64:
- // CHECK-NEXT: ins.d v0[1], x0
- // CHECK-NEXT: ret
return vsetq_lane_u64(a, b, 1);
}
Modified: cfe/trunk/test/CodeGen/aarch64-poly128.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/aarch64-poly128.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/aarch64-poly128.c (original)
+++ cfe/trunk/test/CodeGen/aarch64-poly128.c Wed Mar 9 12:54:42 2016
@@ -1,7 +1,7 @@
// REQUIRES: aarch64-registered-target
// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon \
-// RUN: -ffp-contract=fast -S -O3 -o - %s | FileCheck %s --check-prefix=CHECK \
-// RUN: --check-prefix=CHECK-ARM64
+// RUN: -ffp-contract=fast -emit-llvm -o - %s | opt -S -mem2reg \
+// RUN: | FileCheck %s
// Test new aarch64 intrinsics with poly128
// FIXME: Currently, poly128_t equals to uint128, which will be spilt into
@@ -12,192 +12,238 @@
#include <arm_neon.h>
+// CHECK-LABEL: define void @test_vstrq_p128(i128* %ptr, i128 %val) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128* %ptr to i8*
+// CHECK: [[TMP1:%.*]] = bitcast i8* [[TMP0]] to i128*
+// CHECK: store i128 %val, i128* [[TMP1]]
+// CHECK: ret void
void test_vstrq_p128(poly128_t * ptr, poly128_t val) {
- // CHECK-LABEL: test_vstrq_p128
vstrq_p128(ptr, val);
- // CHECK-ARM64: stp {{x[0-9]+}}, {{x[0-9]+}}, [x0]
}
+// CHECK-LABEL: define i128 @test_vldrq_p128(i128* %ptr) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128* %ptr to i8*
+// CHECK: [[TMP1:%.*]] = bitcast i8* [[TMP0]] to i128*
+// CHECK: [[TMP2:%.*]] = load i128, i128* [[TMP1]]
+// CHECK: ret i128 [[TMP2]]
poly128_t test_vldrq_p128(poly128_t * ptr) {
- // CHECK-LABEL: test_vldrq_p128
return vldrq_p128(ptr);
- // CHECK-ARM64: ldp {{x[0-9]+}}, {{x[0-9]+}}, [x0]
}
+// CHECK-LABEL: define void @test_ld_st_p128(i128* %ptr) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128* %ptr to i8*
+// CHECK: [[TMP1:%.*]] = bitcast i8* [[TMP0]] to i128*
+// CHECK: [[TMP2:%.*]] = load i128, i128* [[TMP1]]
+// CHECK: [[ADD_PTR:%.*]] = getelementptr inbounds i128, i128* %ptr, i64 1
+// CHECK: [[TMP3:%.*]] = bitcast i128* [[ADD_PTR]] to i8*
+// CHECK: [[TMP4:%.*]] = bitcast i8* [[TMP3]] to i128*
+// CHECK: store i128 [[TMP2]], i128* [[TMP4]]
+// CHECK: ret void
void test_ld_st_p128(poly128_t * ptr) {
- // CHECK-LABEL: test_ld_st_p128
vstrq_p128(ptr+1, vldrq_p128(ptr));
- // CHECK-ARM64: ldp [[PLO:x[0-9]+]], [[PHI:x[0-9]+]], [{{x[0-9]+}}]
- // CHECK-ARM64-NEXT: stp [[PLO]], [[PHI]], [{{x[0-9]+}}, #16]
}
+// CHECK-LABEL: define i128 @test_vmull_p64(i64 %a, i64 %b) #0 {
+// CHECK: [[VMULL_P64_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.pmull64(i64 %a, i64 %b) #2
+// CHECK: [[VMULL_P641_I:%.*]] = bitcast <16 x i8> [[VMULL_P64_I]] to i128
+// CHECK: ret i128 [[VMULL_P641_I]]
poly128_t test_vmull_p64(poly64_t a, poly64_t b) {
- // CHECK-LABEL: test_vmull_p64
return vmull_p64(a, b);
- // CHECK: pmull {{v[0-9]+}}.1q, {{v[0-9]+}}.1d, {{v[0-9]+}}.1d
}
+// CHECK-LABEL: define i128 @test_vmull_high_p64(<2 x i64> %a, <2 x i64> %b) #0 {
+// CHECK: [[SHUFFLE_I_I:%.*]] = shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> <i32 1>
+// CHECK: [[TMP0:%.*]] = bitcast <1 x i64> [[SHUFFLE_I_I]] to i64
+// CHECK: [[SHUFFLE_I7_I:%.*]] = shufflevector <2 x i64> %b, <2 x i64> %b, <1 x i32> <i32 1>
+// CHECK: [[TMP1:%.*]] = bitcast <1 x i64> [[SHUFFLE_I7_I]] to i64
+// CHECK: [[VMULL_P64_I_I:%.*]] = call <16 x i8> @llvm.aarch64.neon.pmull64(i64 [[TMP0]], i64 [[TMP1]]) #2
+// CHECK: [[VMULL_P641_I_I:%.*]] = bitcast <16 x i8> [[VMULL_P64_I_I]] to i128
+// CHECK: ret i128 [[VMULL_P641_I_I]]
poly128_t test_vmull_high_p64(poly64x2_t a, poly64x2_t b) {
- // CHECK-LABEL: test_vmull_high_p64
return vmull_high_p64(a, b);
- // CHECK: pmull2 {{v[0-9]+}}.1q, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
}
-// CHECK-LABEL: test_vreinterpretq_p128_s8
-// CHECK: ret
+// CHECK-LABEL: define i128 @test_vreinterpretq_p128_s8(<16 x i8> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <16 x i8> %a to i128
+// CHECK: ret i128 [[TMP0]]
poly128_t test_vreinterpretq_p128_s8(int8x16_t a) {
return vreinterpretq_p128_s8(a);
}
-// CHECK-LABEL: test_vreinterpretq_p128_s16
-// CHECK: ret
+// CHECK-LABEL: define i128 @test_vreinterpretq_p128_s16(<8 x i16> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to i128
+// CHECK: ret i128 [[TMP0]]
poly128_t test_vreinterpretq_p128_s16(int16x8_t a) {
return vreinterpretq_p128_s16(a);
}
-// CHECK-LABEL: test_vreinterpretq_p128_s32
-// CHECK: ret
+// CHECK-LABEL: define i128 @test_vreinterpretq_p128_s32(<4 x i32> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to i128
+// CHECK: ret i128 [[TMP0]]
poly128_t test_vreinterpretq_p128_s32(int32x4_t a) {
return vreinterpretq_p128_s32(a);
}
-// CHECK-LABEL: test_vreinterpretq_p128_s64
-// CHECK: ret
+// CHECK-LABEL: define i128 @test_vreinterpretq_p128_s64(<2 x i64> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to i128
+// CHECK: ret i128 [[TMP0]]
poly128_t test_vreinterpretq_p128_s64(int64x2_t a) {
return vreinterpretq_p128_s64(a);
}
-// CHECK-LABEL: test_vreinterpretq_p128_u8
-// CHECK: ret
+// CHECK-LABEL: define i128 @test_vreinterpretq_p128_u8(<16 x i8> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <16 x i8> %a to i128
+// CHECK: ret i128 [[TMP0]]
poly128_t test_vreinterpretq_p128_u8(uint8x16_t a) {
return vreinterpretq_p128_u8(a);
}
-// CHECK-LABEL: test_vreinterpretq_p128_u16
-// CHECK: ret
+// CHECK-LABEL: define i128 @test_vreinterpretq_p128_u16(<8 x i16> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to i128
+// CHECK: ret i128 [[TMP0]]
poly128_t test_vreinterpretq_p128_u16(uint16x8_t a) {
return vreinterpretq_p128_u16(a);
}
-// CHECK-LABEL: test_vreinterpretq_p128_u32
-// CHECK: ret
+// CHECK-LABEL: define i128 @test_vreinterpretq_p128_u32(<4 x i32> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a to i128
+// CHECK: ret i128 [[TMP0]]
poly128_t test_vreinterpretq_p128_u32(uint32x4_t a) {
return vreinterpretq_p128_u32(a);
}
-// CHECK-LABEL: test_vreinterpretq_p128_u64
-// CHECK: ret
+// CHECK-LABEL: define i128 @test_vreinterpretq_p128_u64(<2 x i64> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to i128
+// CHECK: ret i128 [[TMP0]]
poly128_t test_vreinterpretq_p128_u64(uint64x2_t a) {
return vreinterpretq_p128_u64(a);
}
-// CHECK-LABEL: test_vreinterpretq_p128_f32
-// CHECK: ret
+// CHECK-LABEL: define i128 @test_vreinterpretq_p128_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to i128
+// CHECK: ret i128 [[TMP0]]
poly128_t test_vreinterpretq_p128_f32(float32x4_t a) {
return vreinterpretq_p128_f32(a);
}
-// CHECK-LABEL: test_vreinterpretq_p128_f64
-// CHECK: ret
+// CHECK-LABEL: define i128 @test_vreinterpretq_p128_f64(<2 x double> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x double> %a to i128
+// CHECK: ret i128 [[TMP0]]
poly128_t test_vreinterpretq_p128_f64(float64x2_t a) {
return vreinterpretq_p128_f64(a);
}
-// CHECK-LABEL: test_vreinterpretq_p128_p8
-// CHECK: ret
+// CHECK-LABEL: define i128 @test_vreinterpretq_p128_p8(<16 x i8> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <16 x i8> %a to i128
+// CHECK: ret i128 [[TMP0]]
poly128_t test_vreinterpretq_p128_p8(poly8x16_t a) {
return vreinterpretq_p128_p8(a);
}
-// CHECK-LABEL: test_vreinterpretq_p128_p16
-// CHECK: ret
+// CHECK-LABEL: define i128 @test_vreinterpretq_p128_p16(<8 x i16> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a to i128
+// CHECK: ret i128 [[TMP0]]
poly128_t test_vreinterpretq_p128_p16(poly16x8_t a) {
return vreinterpretq_p128_p16(a);
}
-// CHECK-LABEL: test_vreinterpretq_p128_p64
-// CHECK: ret
+// CHECK-LABEL: define i128 @test_vreinterpretq_p128_p64(<2 x i64> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to i128
+// CHECK: ret i128 [[TMP0]]
poly128_t test_vreinterpretq_p128_p64(poly64x2_t a) {
return vreinterpretq_p128_p64(a);
}
-// CHECK-LABEL: test_vreinterpretq_s8_p128
-// CHECK: ret
+// CHECK-LABEL: define <16 x i8> @test_vreinterpretq_s8_p128(i128 %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128 %a to <16 x i8>
+// CHECK: ret <16 x i8> [[TMP0]]
int8x16_t test_vreinterpretq_s8_p128(poly128_t a) {
return vreinterpretq_s8_p128(a);
}
-// CHECK-LABEL: test_vreinterpretq_s16_p128
-// CHECK: ret
+// CHECK-LABEL: define <8 x i16> @test_vreinterpretq_s16_p128(i128 %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128 %a to <8 x i16>
+// CHECK: ret <8 x i16> [[TMP0]]
int16x8_t test_vreinterpretq_s16_p128(poly128_t a) {
return vreinterpretq_s16_p128(a);
}
-// CHECK-LABEL: test_vreinterpretq_s32_p128
-// CHECK: ret
+// CHECK-LABEL: define <4 x i32> @test_vreinterpretq_s32_p128(i128 %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128 %a to <4 x i32>
+// CHECK: ret <4 x i32> [[TMP0]]
int32x4_t test_vreinterpretq_s32_p128(poly128_t a) {
return vreinterpretq_s32_p128(a);
}
-// CHECK-LABEL: test_vreinterpretq_s64_p128
-// CHECK: ret
+// CHECK-LABEL: define <2 x i64> @test_vreinterpretq_s64_p128(i128 %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128 %a to <2 x i64>
+// CHECK: ret <2 x i64> [[TMP0]]
int64x2_t test_vreinterpretq_s64_p128(poly128_t a) {
return vreinterpretq_s64_p128(a);
}
-// CHECK-LABEL: test_vreinterpretq_u8_p128
-// CHECK: ret
+// CHECK-LABEL: define <16 x i8> @test_vreinterpretq_u8_p128(i128 %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128 %a to <16 x i8>
+// CHECK: ret <16 x i8> [[TMP0]]
uint8x16_t test_vreinterpretq_u8_p128(poly128_t a) {
return vreinterpretq_u8_p128(a);
}
-// CHECK-LABEL: test_vreinterpretq_u16_p128
-// CHECK: ret
+// CHECK-LABEL: define <8 x i16> @test_vreinterpretq_u16_p128(i128 %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128 %a to <8 x i16>
+// CHECK: ret <8 x i16> [[TMP0]]
uint16x8_t test_vreinterpretq_u16_p128(poly128_t a) {
return vreinterpretq_u16_p128(a);
}
-// CHECK-LABEL: test_vreinterpretq_u32_p128
-// CHECK: ret
+// CHECK-LABEL: define <4 x i32> @test_vreinterpretq_u32_p128(i128 %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128 %a to <4 x i32>
+// CHECK: ret <4 x i32> [[TMP0]]
uint32x4_t test_vreinterpretq_u32_p128(poly128_t a) {
return vreinterpretq_u32_p128(a);
}
-// CHECK-LABEL: test_vreinterpretq_u64_p128
-// CHECK: ret
+// CHECK-LABEL: define <2 x i64> @test_vreinterpretq_u64_p128(i128 %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128 %a to <2 x i64>
+// CHECK: ret <2 x i64> [[TMP0]]
uint64x2_t test_vreinterpretq_u64_p128(poly128_t a) {
return vreinterpretq_u64_p128(a);
}
-// CHECK-LABEL: test_vreinterpretq_f32_p128
-// CHECK: ret
+// CHECK-LABEL: define <4 x float> @test_vreinterpretq_f32_p128(i128 %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128 %a to <4 x float>
+// CHECK: ret <4 x float> [[TMP0]]
float32x4_t test_vreinterpretq_f32_p128(poly128_t a) {
return vreinterpretq_f32_p128(a);
}
-// CHECK-LABEL: test_vreinterpretq_f64_p128
-// CHECK: ret
+// CHECK-LABEL: define <2 x double> @test_vreinterpretq_f64_p128(i128 %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128 %a to <2 x double>
+// CHECK: ret <2 x double> [[TMP0]]
float64x2_t test_vreinterpretq_f64_p128(poly128_t a) {
return vreinterpretq_f64_p128(a);
}
-// CHECK-LABEL: test_vreinterpretq_p8_p128
-// CHECK: ret
+// CHECK-LABEL: define <16 x i8> @test_vreinterpretq_p8_p128(i128 %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128 %a to <16 x i8>
+// CHECK: ret <16 x i8> [[TMP0]]
poly8x16_t test_vreinterpretq_p8_p128(poly128_t a) {
return vreinterpretq_p8_p128(a);
}
-// CHECK-LABEL: test_vreinterpretq_p16_p128
-// CHECK: ret
+// CHECK-LABEL: define <8 x i16> @test_vreinterpretq_p16_p128(i128 %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128 %a to <8 x i16>
+// CHECK: ret <8 x i16> [[TMP0]]
poly16x8_t test_vreinterpretq_p16_p128(poly128_t a) {
return vreinterpretq_p16_p128(a);
}
-// CHECK-LABEL: test_vreinterpretq_p64_p128
-// CHECK: ret
+// CHECK-LABEL: define <2 x i64> @test_vreinterpretq_p64_p128(i128 %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i128 %a to <2 x i64>
+// CHECK: ret <2 x i64> [[TMP0]]
poly64x2_t test_vreinterpretq_p64_p128(poly128_t a) {
return vreinterpretq_p64_p128(a);
}
Modified: cfe/trunk/test/CodeGen/aarch64-poly64.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/aarch64-poly64.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/aarch64-poly64.c (original)
+++ cfe/trunk/test/CodeGen/aarch64-poly64.c Wed Mar 9 12:54:42 2016
@@ -1,299 +1,634 @@
-// FIXME: This is a front-end test that depends on LLVM optimizations (-O3).
-// It should be split into separate files for front/middle/back-end testing.
-
-// REQUIRES: aarch64-registered-target
// RUN: %clang_cc1 -triple arm64-none-linux-gnu -target-feature +neon \
-// RUN: -ffp-contract=fast -S -O3 -o - %s | FileCheck %s --check-prefix=CHECK \
-// RUN: --check-prefix=CHECK-ARM64
+// RUN: -ffp-contract=fast -emit-llvm -o - %s | opt -S -mem2reg \
+// RUN: | FileCheck %s
// Test new aarch64 intrinsics with poly64
#include <arm_neon.h>
+// CHECK-LABEL: define <1 x i64> @test_vceq_p64(<1 x i64> %a, <1 x i64> %b) #0 {
+// CHECK: [[CMP_I:%.*]] = icmp eq <1 x i64> %a, %b
+// CHECK: [[SEXT_I:%.*]] = sext <1 x i1> [[CMP_I]] to <1 x i64>
+// CHECK: ret <1 x i64> [[SEXT_I]]
uint64x1_t test_vceq_p64(poly64x1_t a, poly64x1_t b) {
- // CHECK-LABEL: test_vceq_p64
return vceq_p64(a, b);
- // CHECK: cmeq {{d[0-9]+}}, {{d[0-9]+}}, {{d[0-9]+}}
}
+// CHECK-LABEL: define <2 x i64> @test_vceqq_p64(<2 x i64> %a, <2 x i64> %b) #0 {
+// CHECK: [[CMP_I:%.*]] = icmp eq <2 x i64> %a, %b
+// CHECK: [[SEXT_I:%.*]] = sext <2 x i1> [[CMP_I]] to <2 x i64>
+// CHECK: ret <2 x i64> [[SEXT_I]]
uint64x2_t test_vceqq_p64(poly64x2_t a, poly64x2_t b) {
- // CHECK-LABEL: test_vceqq_p64
return vceqq_p64(a, b);
- // CHECK: cmeq {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
}
+// CHECK-LABEL: define <1 x i64> @test_vtst_p64(<1 x i64> %a, <1 x i64> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <1 x i64> %b to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP1]] to <1 x i64>
+// CHECK: [[TMP4:%.*]] = and <1 x i64> [[TMP2]], [[TMP3]]
+// CHECK: [[TMP5:%.*]] = icmp ne <1 x i64> [[TMP4]], zeroinitializer
+// CHECK: [[VTST_I:%.*]] = sext <1 x i1> [[TMP5]] to <1 x i64>
+// CHECK: ret <1 x i64> [[VTST_I]]
uint64x1_t test_vtst_p64(poly64x1_t a, poly64x1_t b) {
- // CHECK-LABEL: test_vtst_p64
return vtst_p64(a, b);
- // CHECK: cmtst {{d[0-9]+}}, {{d[0-9]+}}, {{d[0-9]+}}
}
+// CHECK-LABEL: define <2 x i64> @test_vtstq_p64(<2 x i64> %a, <2 x i64> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <2 x i64> %b to <16 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP1]] to <2 x i64>
+// CHECK: [[TMP4:%.*]] = and <2 x i64> [[TMP2]], [[TMP3]]
+// CHECK: [[TMP5:%.*]] = icmp ne <2 x i64> [[TMP4]], zeroinitializer
+// CHECK: [[VTST_I:%.*]] = sext <2 x i1> [[TMP5]] to <2 x i64>
+// CHECK: ret <2 x i64> [[VTST_I]]
uint64x2_t test_vtstq_p64(poly64x2_t a, poly64x2_t b) {
- // CHECK-LABEL: test_vtstq_p64
return vtstq_p64(a, b);
- // CHECK: cmtst {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
}
+// CHECK-LABEL: define <1 x i64> @test_vbsl_p64(<1 x i64> %a, <1 x i64> %b, <1 x i64> %c) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <1 x i64> %b to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <1 x i64> %c to <8 x i8>
+// CHECK: [[VBSL_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64>
+// CHECK: [[VBSL1_I:%.*]] = bitcast <8 x i8> [[TMP1]] to <1 x i64>
+// CHECK: [[VBSL2_I:%.*]] = bitcast <8 x i8> [[TMP2]] to <1 x i64>
+// CHECK: [[VBSL3_I:%.*]] = and <1 x i64> [[VBSL_I]], [[VBSL1_I]]
+// CHECK: [[TMP3:%.*]] = xor <1 x i64> [[VBSL_I]], <i64 -1>
+// CHECK: [[VBSL4_I:%.*]] = and <1 x i64> [[TMP3]], [[VBSL2_I]]
+// CHECK: [[VBSL5_I:%.*]] = or <1 x i64> [[VBSL3_I]], [[VBSL4_I]]
+// CHECK: ret <1 x i64> [[VBSL5_I]]
poly64x1_t test_vbsl_p64(poly64x1_t a, poly64x1_t b, poly64x1_t c) {
- // CHECK-LABEL: test_vbsl_p64
return vbsl_p64(a, b, c);
- // CHECK: bsl {{v[0-9]+}}.8b, {{v[0-9]+}}.8b, {{v[0-9]+}}.8b
}
+// CHECK-LABEL: define <2 x i64> @test_vbslq_p64(<2 x i64> %a, <2 x i64> %b, <2 x i64> %c) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <2 x i64> %b to <16 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <2 x i64> %c to <16 x i8>
+// CHECK: [[VBSL_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64>
+// CHECK: [[VBSL1_I:%.*]] = bitcast <16 x i8> [[TMP1]] to <2 x i64>
+// CHECK: [[VBSL2_I:%.*]] = bitcast <16 x i8> [[TMP2]] to <2 x i64>
+// CHECK: [[VBSL3_I:%.*]] = and <2 x i64> [[VBSL_I]], [[VBSL1_I]]
+// CHECK: [[TMP3:%.*]] = xor <2 x i64> [[VBSL_I]], <i64 -1, i64 -1>
+// CHECK: [[VBSL4_I:%.*]] = and <2 x i64> [[TMP3]], [[VBSL2_I]]
+// CHECK: [[VBSL5_I:%.*]] = or <2 x i64> [[VBSL3_I]], [[VBSL4_I]]
+// CHECK: ret <2 x i64> [[VBSL5_I]]
poly64x2_t test_vbslq_p64(poly64x2_t a, poly64x2_t b, poly64x2_t c) {
- // CHECK-LABEL: test_vbslq_p64
return vbslq_p64(a, b, c);
- // CHECK: bsl {{v[0-9]+}}.16b, {{v[0-9]+}}.16b, {{v[0-9]+}}.16b
}
+// CHECK-LABEL: define i64 @test_vget_lane_p64(<1 x i64> %v) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %v to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <1 x i64> [[TMP1]], i32 0
+// CHECK: ret i64 [[VGET_LANE]]
poly64_t test_vget_lane_p64(poly64x1_t v) {
- // CHECK-LABEL: test_vget_lane_p64
return vget_lane_p64(v, 0);
- // CHECK: fmov {{x[0-9]+}}, {{d[0-9]+}}
}
+// CHECK-LABEL: define i64 @test_vgetq_lane_p64(<2 x i64> %v) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %v to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <2 x i64> [[TMP1]], i32 1
+// CHECK: ret i64 [[VGETQ_LANE]]
poly64_t test_vgetq_lane_p64(poly64x2_t v) {
- // CHECK-LABEL: test_vgetq_lane_p64
return vgetq_lane_p64(v, 1);
- // CHECK: {{mov|umov}} {{x[0-9]+}}, {{v[0-9]+}}.d[1]
}
+// CHECK-LABEL: define <1 x i64> @test_vset_lane_p64(i64 %a, <1 x i64> %v) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %v to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <1 x i64> [[TMP1]], i64 %a, i32 0
+// CHECK: ret <1 x i64> [[VSET_LANE]]
poly64x1_t test_vset_lane_p64(poly64_t a, poly64x1_t v) {
- // CHECK-LABEL: test_vset_lane_p64
return vset_lane_p64(a, v, 0);
- // CHECK: fmov {{d[0-9]+}}, {{x[0-9]+}}
}
+// CHECK-LABEL: define <2 x i64> @test_vsetq_lane_p64(i64 %a, <2 x i64> %v) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %v to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <2 x i64> [[TMP1]], i64 %a, i32 1
+// CHECK: ret <2 x i64> [[VSET_LANE]]
poly64x2_t test_vsetq_lane_p64(poly64_t a, poly64x2_t v) {
- // CHECK-LABEL: test_vsetq_lane_p64
return vsetq_lane_p64(a, v, 1);
- // CHECK: ins {{v[0-9]+}}.d[1], {{x[0-9]+}}
}
+// CHECK-LABEL: define <1 x i64> @test_vcopy_lane_p64(<1 x i64> %a, <1 x i64> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <1 x i64> [[TMP1]], i32 0
+// CHECK: [[TMP2:%.*]] = bitcast <1 x i64> %a to <8 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP2]] to <1 x i64>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <1 x i64> [[TMP3]], i64 [[VGET_LANE]], i32 0
+// CHECK: ret <1 x i64> [[VSET_LANE]]
poly64x1_t test_vcopy_lane_p64(poly64x1_t a, poly64x1_t b) {
- // CHECK-LABEL: test_vcopy_lane_p64
return vcopy_lane_p64(a, 0, b, 0);
- // CHECK-ARM64: mov v0.16b, v1.16b
}
+// CHECK-LABEL: define <2 x i64> @test_vcopyq_lane_p64(<2 x i64> %a, <1 x i64> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %b to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64>
+// CHECK: [[VGET_LANE:%.*]] = extractelement <1 x i64> [[TMP1]], i32 0
+// CHECK: [[TMP2:%.*]] = bitcast <2 x i64> %a to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP2]] to <2 x i64>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <2 x i64> [[TMP3]], i64 [[VGET_LANE]], i32 1
+// CHECK: ret <2 x i64> [[VSET_LANE]]
poly64x2_t test_vcopyq_lane_p64(poly64x2_t a, poly64x1_t b) {
- // CHECK-LABEL: test_vcopyq_lane_p64
return vcopyq_lane_p64(a, 1, b, 0);
- // CHECK: zip1 v0.2d, v0.2d, v1.2d
}
+// CHECK-LABEL: define <2 x i64> @test_vcopyq_laneq_p64(<2 x i64> %a, <2 x i64> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %b to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <2 x i64> [[TMP1]], i32 1
+// CHECK: [[TMP2:%.*]] = bitcast <2 x i64> %a to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP2]] to <2 x i64>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <2 x i64> [[TMP3]], i64 [[VGETQ_LANE]], i32 1
+// CHECK: ret <2 x i64> [[VSET_LANE]]
poly64x2_t test_vcopyq_laneq_p64(poly64x2_t a, poly64x2_t b) {
- // CHECK-LABEL: test_vcopyq_laneq_p64
return vcopyq_laneq_p64(a, 1, b, 1);
}
+// CHECK-LABEL: define <1 x i64> @test_vcreate_p64(i64 %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i64 %a to <1 x i64>
+// CHECK: ret <1 x i64> [[TMP0]]
poly64x1_t test_vcreate_p64(uint64_t a) {
- // CHECK-LABEL: test_vcreate_p64
return vcreate_p64(a);
- // CHECK: fmov {{d[0-9]+}}, {{x[0-9]+}}
}
+// CHECK-LABEL: define <1 x i64> @test_vdup_n_p64(i64 %a) #0 {
+// CHECK: [[VECINIT_I:%.*]] = insertelement <1 x i64> undef, i64 %a, i32 0
+// CHECK: ret <1 x i64> [[VECINIT_I]]
poly64x1_t test_vdup_n_p64(poly64_t a) {
- // CHECK-LABEL: test_vdup_n_p64
return vdup_n_p64(a);
- // CHECK: fmov {{d[0-9]+}}, {{x[0-9]+}}
}
+// CHECK-LABEL: define <2 x i64> @test_vdupq_n_p64(i64 %a) #0 {
+// CHECK: [[VECINIT_I:%.*]] = insertelement <2 x i64> undef, i64 %a, i32 0
+// CHECK: [[VECINIT1_I:%.*]] = insertelement <2 x i64> [[VECINIT_I]], i64 %a, i32 1
+// CHECK: ret <2 x i64> [[VECINIT1_I]]
poly64x2_t test_vdupq_n_p64(poly64_t a) {
- // CHECK-LABEL: test_vdupq_n_p64
return vdupq_n_p64(a);
- // CHECK: dup {{v[0-9]+}}.2d, {{x[0-9]+}}
}
+// CHECK-LABEL: define <1 x i64> @test_vmov_n_p64(i64 %a) #0 {
+// CHECK: [[VECINIT_I:%.*]] = insertelement <1 x i64> undef, i64 %a, i32 0
+// CHECK: ret <1 x i64> [[VECINIT_I]]
poly64x1_t test_vmov_n_p64(poly64_t a) {
- // CHECK-LABEL: test_vmov_n_p64
return vmov_n_p64(a);
- // CHECK: fmov {{d[0-9]+}}, {{x[0-9]+}}
}
+// CHECK-LABEL: define <2 x i64> @test_vmovq_n_p64(i64 %a) #0 {
+// CHECK: [[VECINIT_I:%.*]] = insertelement <2 x i64> undef, i64 %a, i32 0
+// CHECK: [[VECINIT1_I:%.*]] = insertelement <2 x i64> [[VECINIT_I]], i64 %a, i32 1
+// CHECK: ret <2 x i64> [[VECINIT1_I]]
poly64x2_t test_vmovq_n_p64(poly64_t a) {
- // CHECK-LABEL: test_vmovq_n_p64
return vmovq_n_p64(a);
- // CHECK: dup {{v[0-9]+}}.2d, {{x[0-9]+}}
}
+// CHECK-LABEL: define <1 x i64> @test_vdup_lane_p64(<1 x i64> %vec) #0 {
+// CHECK: [[SHUFFLE:%.*]] = shufflevector <1 x i64> %vec, <1 x i64> %vec, <1 x i32> zeroinitializer
+// CHECK: ret <1 x i64> [[SHUFFLE]]
poly64x1_t test_vdup_lane_p64(poly64x1_t vec) {
- // CHECK-LABEL: test_vdup_lane_p64
return vdup_lane_p64(vec, 0);
- // CHECK: ret
}
+// CHECK-LABEL: define <2 x i64> @test_vdupq_lane_p64(<1 x i64> %vec) #0 {
+// CHECK: [[SHUFFLE:%.*]] = shufflevector <1 x i64> %vec, <1 x i64> %vec, <2 x i32> zeroinitializer
+// CHECK: ret <2 x i64> [[SHUFFLE]]
poly64x2_t test_vdupq_lane_p64(poly64x1_t vec) {
- // CHECK-LABEL: test_vdupq_lane_p64
return vdupq_lane_p64(vec, 0);
- // CHECK: dup {{v[0-9]+}}.2d, {{v[0-9]+}}.d[0]
}
+// CHECK-LABEL: define <2 x i64> @test_vdupq_laneq_p64(<2 x i64> %vec) #0 {
+// CHECK: [[SHUFFLE:%.*]] = shufflevector <2 x i64> %vec, <2 x i64> %vec, <2 x i32> <i32 1, i32 1>
+// CHECK: ret <2 x i64> [[SHUFFLE]]
poly64x2_t test_vdupq_laneq_p64(poly64x2_t vec) {
- // CHECK-LABEL: test_vdupq_laneq_p64
return vdupq_laneq_p64(vec, 1);
- // CHECK: dup {{v[0-9]+}}.2d, {{v[0-9]+}}.d[1]
}
+// CHECK-LABEL: define <2 x i64> @test_vcombine_p64(<1 x i64> %low, <1 x i64> %high) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <1 x i64> %low, <1 x i64> %high, <2 x i32> <i32 0, i32 1>
+// CHECK: ret <2 x i64> [[SHUFFLE_I]]
poly64x2_t test_vcombine_p64(poly64x1_t low, poly64x1_t high) {
- // CHECK-LABEL: test_vcombine_p64
return vcombine_p64(low, high);
- // CHECK: ins {{v[0-9]+}}.d[1], {{v[0-9]+}}.d[0]
}
+// CHECK-LABEL: define <1 x i64> @test_vld1_p64(i64* %ptr) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[TMP1:%.*]] = bitcast i8* [[TMP0]] to <1 x i64>*
+// CHECK: [[TMP2:%.*]] = load <1 x i64>, <1 x i64>* [[TMP1]]
+// CHECK: ret <1 x i64> [[TMP2]]
poly64x1_t test_vld1_p64(poly64_t const * ptr) {
- // CHECK-LABEL: test_vld1_p64
return vld1_p64(ptr);
- // CHECK-ARM64: ldr {{d[0-9]+}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define <2 x i64> @test_vld1q_p64(i64* %ptr) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[TMP1:%.*]] = bitcast i8* [[TMP0]] to <2 x i64>*
+// CHECK: [[TMP2:%.*]] = load <2 x i64>, <2 x i64>* [[TMP1]]
+// CHECK: ret <2 x i64> [[TMP2]]
poly64x2_t test_vld1q_p64(poly64_t const * ptr) {
- // CHECK-LABEL: test_vld1q_p64
return vld1q_p64(ptr);
- // CHECK-ARM64: ldr {{q[0-9]+}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define void @test_vst1_p64(i64* %ptr, <1 x i64> %val) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[TMP1:%.*]] = bitcast <1 x i64> %val to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast i8* [[TMP0]] to <1 x i64>*
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP1]] to <1 x i64>
+// CHECK: store <1 x i64> [[TMP3]], <1 x i64>* [[TMP2]]
+// CHECK: ret void
void test_vst1_p64(poly64_t * ptr, poly64x1_t val) {
- // CHECK-LABEL: test_vst1_p64
return vst1_p64(ptr, val);
- // CHECK-ARM64: str {{d[0-9]+}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define void @test_vst1q_p64(i64* %ptr, <2 x i64> %val) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[TMP1:%.*]] = bitcast <2 x i64> %val to <16 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast i8* [[TMP0]] to <2 x i64>*
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP1]] to <2 x i64>
+// CHECK: store <2 x i64> [[TMP3]], <2 x i64>* [[TMP2]]
+// CHECK: ret void
void test_vst1q_p64(poly64_t * ptr, poly64x2_t val) {
- // CHECK-LABEL: test_vst1q_p64
return vst1q_p64(ptr, val);
- // CHECK-ARM64: str {{q[0-9]+}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define %struct.poly64x1x2_t @test_vld2_p64(i64* %ptr) #0 {
+// CHECK: [[RETVAL:%.*]] = alloca %struct.poly64x1x2_t, align 8
+// CHECK: [[__RET:%.*]] = alloca %struct.poly64x1x2_t, align 8
+// CHECK: [[TMP0:%.*]] = bitcast %struct.poly64x1x2_t* [[__RET]] to i8*
+// CHECK: [[TMP1:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <1 x i64>*
+// CHECK: [[VLD2:%.*]] = call { <1 x i64>, <1 x i64> } @llvm.aarch64.neon.ld2.v1i64.p0v1i64(<1 x i64>* [[TMP2]])
+// CHECK: [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <1 x i64>, <1 x i64> }*
+// CHECK: store { <1 x i64>, <1 x i64> } [[VLD2]], { <1 x i64>, <1 x i64> }* [[TMP3]]
+// CHECK: [[TMP4:%.*]] = bitcast %struct.poly64x1x2_t* [[RETVAL]] to i8*
+// CHECK: [[TMP5:%.*]] = bitcast %struct.poly64x1x2_t* [[__RET]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 16, i32 8, i1 false)
+// CHECK: [[TMP6:%.*]] = load %struct.poly64x1x2_t, %struct.poly64x1x2_t* [[RETVAL]], align 8
+// CHECK: ret %struct.poly64x1x2_t [[TMP6]]
poly64x1x2_t test_vld2_p64(poly64_t const * ptr) {
- // CHECK-LABEL: test_vld2_p64
return vld2_p64(ptr);
- // CHECK: ld1 {{{ *v[0-9]+.1d, v[0-9]+.1d *}}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define %struct.poly64x2x2_t @test_vld2q_p64(i64* %ptr) #0 {
+// CHECK: [[RETVAL:%.*]] = alloca %struct.poly64x2x2_t, align 16
+// CHECK: [[__RET:%.*]] = alloca %struct.poly64x2x2_t, align 16
+// CHECK: [[TMP0:%.*]] = bitcast %struct.poly64x2x2_t* [[__RET]] to i8*
+// CHECK: [[TMP1:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <2 x i64>*
+// CHECK: [[VLD2:%.*]] = call { <2 x i64>, <2 x i64> } @llvm.aarch64.neon.ld2.v2i64.p0v2i64(<2 x i64>* [[TMP2]])
+// CHECK: [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <2 x i64>, <2 x i64> }*
+// CHECK: store { <2 x i64>, <2 x i64> } [[VLD2]], { <2 x i64>, <2 x i64> }* [[TMP3]]
+// CHECK: [[TMP4:%.*]] = bitcast %struct.poly64x2x2_t* [[RETVAL]] to i8*
+// CHECK: [[TMP5:%.*]] = bitcast %struct.poly64x2x2_t* [[__RET]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 32, i32 16, i1 false)
+// CHECK: [[TMP6:%.*]] = load %struct.poly64x2x2_t, %struct.poly64x2x2_t* [[RETVAL]], align 16
+// CHECK: ret %struct.poly64x2x2_t [[TMP6]]
poly64x2x2_t test_vld2q_p64(poly64_t const * ptr) {
- // CHECK-LABEL: test_vld2q_p64
return vld2q_p64(ptr);
- // CHECK: ld2 {{{ *v[0-9]+.2d, v[0-9]+.2d *}}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define %struct.poly64x1x3_t @test_vld3_p64(i64* %ptr) #0 {
+// CHECK: [[RETVAL:%.*]] = alloca %struct.poly64x1x3_t, align 8
+// CHECK: [[__RET:%.*]] = alloca %struct.poly64x1x3_t, align 8
+// CHECK: [[TMP0:%.*]] = bitcast %struct.poly64x1x3_t* [[__RET]] to i8*
+// CHECK: [[TMP1:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <1 x i64>*
+// CHECK: [[VLD3:%.*]] = call { <1 x i64>, <1 x i64>, <1 x i64> } @llvm.aarch64.neon.ld3.v1i64.p0v1i64(<1 x i64>* [[TMP2]])
+// CHECK: [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <1 x i64>, <1 x i64>, <1 x i64> }*
+// CHECK: store { <1 x i64>, <1 x i64>, <1 x i64> } [[VLD3]], { <1 x i64>, <1 x i64>, <1 x i64> }* [[TMP3]]
+// CHECK: [[TMP4:%.*]] = bitcast %struct.poly64x1x3_t* [[RETVAL]] to i8*
+// CHECK: [[TMP5:%.*]] = bitcast %struct.poly64x1x3_t* [[__RET]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 24, i32 8, i1 false)
+// CHECK: [[TMP6:%.*]] = load %struct.poly64x1x3_t, %struct.poly64x1x3_t* [[RETVAL]], align 8
+// CHECK: ret %struct.poly64x1x3_t [[TMP6]]
poly64x1x3_t test_vld3_p64(poly64_t const * ptr) {
- // CHECK-LABEL: test_vld3_p64
return vld3_p64(ptr);
- // CHECK: ld1 {{{ *v[0-9]+.1d, v[0-9]+.1d, v[0-9]+.1d *}}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define %struct.poly64x2x3_t @test_vld3q_p64(i64* %ptr) #0 {
+// CHECK: [[RETVAL:%.*]] = alloca %struct.poly64x2x3_t, align 16
+// CHECK: [[__RET:%.*]] = alloca %struct.poly64x2x3_t, align 16
+// CHECK: [[TMP0:%.*]] = bitcast %struct.poly64x2x3_t* [[__RET]] to i8*
+// CHECK: [[TMP1:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <2 x i64>*
+// CHECK: [[VLD3:%.*]] = call { <2 x i64>, <2 x i64>, <2 x i64> } @llvm.aarch64.neon.ld3.v2i64.p0v2i64(<2 x i64>* [[TMP2]])
+// CHECK: [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <2 x i64>, <2 x i64>, <2 x i64> }*
+// CHECK: store { <2 x i64>, <2 x i64>, <2 x i64> } [[VLD3]], { <2 x i64>, <2 x i64>, <2 x i64> }* [[TMP3]]
+// CHECK: [[TMP4:%.*]] = bitcast %struct.poly64x2x3_t* [[RETVAL]] to i8*
+// CHECK: [[TMP5:%.*]] = bitcast %struct.poly64x2x3_t* [[__RET]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 48, i32 16, i1 false)
+// CHECK: [[TMP6:%.*]] = load %struct.poly64x2x3_t, %struct.poly64x2x3_t* [[RETVAL]], align 16
+// CHECK: ret %struct.poly64x2x3_t [[TMP6]]
poly64x2x3_t test_vld3q_p64(poly64_t const * ptr) {
- // CHECK-LABEL: test_vld3q_p64
return vld3q_p64(ptr);
- // CHECK: ld3 {{{ *v[0-9]+.2d, v[0-9]+.2d, v[0-9]+.2d *}}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define %struct.poly64x1x4_t @test_vld4_p64(i64* %ptr) #0 {
+// CHECK: [[RETVAL:%.*]] = alloca %struct.poly64x1x4_t, align 8
+// CHECK: [[__RET:%.*]] = alloca %struct.poly64x1x4_t, align 8
+// CHECK: [[TMP0:%.*]] = bitcast %struct.poly64x1x4_t* [[__RET]] to i8*
+// CHECK: [[TMP1:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <1 x i64>*
+// CHECK: [[VLD4:%.*]] = call { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @llvm.aarch64.neon.ld4.v1i64.p0v1i64(<1 x i64>* [[TMP2]])
+// CHECK: [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> }*
+// CHECK: store { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } [[VLD4]], { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> }* [[TMP3]]
+// CHECK: [[TMP4:%.*]] = bitcast %struct.poly64x1x4_t* [[RETVAL]] to i8*
+// CHECK: [[TMP5:%.*]] = bitcast %struct.poly64x1x4_t* [[__RET]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 32, i32 8, i1 false)
+// CHECK: [[TMP6:%.*]] = load %struct.poly64x1x4_t, %struct.poly64x1x4_t* [[RETVAL]], align 8
+// CHECK: ret %struct.poly64x1x4_t [[TMP6]]
poly64x1x4_t test_vld4_p64(poly64_t const * ptr) {
- // CHECK-LABEL: test_vld4_p64
return vld4_p64(ptr);
- // CHECK: ld1 {{{ *v[0-9]+.1d, v[0-9]+.1d, v[0-9]+.1d, v[0-9]+.1d *}}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define %struct.poly64x2x4_t @test_vld4q_p64(i64* %ptr) #0 {
+// CHECK: [[RETVAL:%.*]] = alloca %struct.poly64x2x4_t, align 16
+// CHECK: [[__RET:%.*]] = alloca %struct.poly64x2x4_t, align 16
+// CHECK: [[TMP0:%.*]] = bitcast %struct.poly64x2x4_t* [[__RET]] to i8*
+// CHECK: [[TMP1:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[TMP2:%.*]] = bitcast i8* [[TMP1]] to <2 x i64>*
+// CHECK: [[VLD4:%.*]] = call { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.aarch64.neon.ld4.v2i64.p0v2i64(<2 x i64>* [[TMP2]])
+// CHECK: [[TMP3:%.*]] = bitcast i8* [[TMP0]] to { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> }*
+// CHECK: store { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } [[VLD4]], { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> }* [[TMP3]]
+// CHECK: [[TMP4:%.*]] = bitcast %struct.poly64x2x4_t* [[RETVAL]] to i8*
+// CHECK: [[TMP5:%.*]] = bitcast %struct.poly64x2x4_t* [[__RET]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP4]], i8* [[TMP5]], i64 64, i32 16, i1 false)
+// CHECK: [[TMP6:%.*]] = load %struct.poly64x2x4_t, %struct.poly64x2x4_t* [[RETVAL]], align 16
+// CHECK: ret %struct.poly64x2x4_t [[TMP6]]
poly64x2x4_t test_vld4q_p64(poly64_t const * ptr) {
- // CHECK-LABEL: test_vld4q_p64
return vld4q_p64(ptr);
- // CHECK: ld4 {{{ *v[0-9]+.2d, v[0-9]+.2d, v[0-9]+.2d, v[0-9]+.2d *}}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define void @test_vst2_p64(i64* %ptr, [2 x <1 x i64>] %val.coerce) #0 {
+// CHECK: [[VAL:%.*]] = alloca %struct.poly64x1x2_t, align 8
+// CHECK: [[__S1:%.*]] = alloca %struct.poly64x1x2_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly64x1x2_t, %struct.poly64x1x2_t* [[VAL]], i32 0, i32 0
+// CHECK: store [2 x <1 x i64>] [[VAL]].coerce, [2 x <1 x i64>]* [[COERCE_DIVE]], align 8
+// CHECK: [[TMP0:%.*]] = bitcast %struct.poly64x1x2_t* [[__S1]] to i8*
+// CHECK: [[TMP1:%.*]] = bitcast %struct.poly64x1x2_t* [[VAL]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP0]], i8* [[TMP1]], i64 16, i32 8, i1 false)
+// CHECK: [[TMP2:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[VAL1:%.*]] = getelementptr inbounds %struct.poly64x1x2_t, %struct.poly64x1x2_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x <1 x i64>], [2 x <1 x i64>]* [[VAL]]1, i64 0, i64 0
+// CHECK: [[TMP3:%.*]] = load <1 x i64>, <1 x i64>* [[ARRAYIDX]], align 8
+// CHECK: [[TMP4:%.*]] = bitcast <1 x i64> [[TMP3]] to <8 x i8>
+// CHECK: [[VAL2:%.*]] = getelementptr inbounds %struct.poly64x1x2_t, %struct.poly64x1x2_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX3:%.*]] = getelementptr inbounds [2 x <1 x i64>], [2 x <1 x i64>]* [[VAL]]2, i64 0, i64 1
+// CHECK: [[TMP5:%.*]] = load <1 x i64>, <1 x i64>* [[ARRAYIDX]]3, align 8
+// CHECK: [[TMP6:%.*]] = bitcast <1 x i64> [[TMP5]] to <8 x i8>
+// CHECK: [[TMP7:%.*]] = bitcast <8 x i8> [[TMP4]] to <1 x i64>
+// CHECK: [[TMP8:%.*]] = bitcast <8 x i8> [[TMP6]] to <1 x i64>
+// CHECK: call void @llvm.aarch64.neon.st2.v1i64.p0i8(<1 x i64> [[TMP7]], <1 x i64> [[TMP8]], i8* [[TMP2]])
+// CHECK: ret void
void test_vst2_p64(poly64_t * ptr, poly64x1x2_t val) {
- // CHECK-LABEL: test_vst2_p64
return vst2_p64(ptr, val);
- // CHECK: st1 {{{ *v[0-9]+.1d, v[0-9]+.1d *}}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define void @test_vst2q_p64(i64* %ptr, [2 x <2 x i64>] %val.coerce) #0 {
+// CHECK: [[VAL:%.*]] = alloca %struct.poly64x2x2_t, align 16
+// CHECK: [[__S1:%.*]] = alloca %struct.poly64x2x2_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly64x2x2_t, %struct.poly64x2x2_t* [[VAL]], i32 0, i32 0
+// CHECK: store [2 x <2 x i64>] [[VAL]].coerce, [2 x <2 x i64>]* [[COERCE_DIVE]], align 16
+// CHECK: [[TMP0:%.*]] = bitcast %struct.poly64x2x2_t* [[__S1]] to i8*
+// CHECK: [[TMP1:%.*]] = bitcast %struct.poly64x2x2_t* [[VAL]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP0]], i8* [[TMP1]], i64 32, i32 16, i1 false)
+// CHECK: [[TMP2:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[VAL1:%.*]] = getelementptr inbounds %struct.poly64x2x2_t, %struct.poly64x2x2_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x <2 x i64>], [2 x <2 x i64>]* [[VAL]]1, i64 0, i64 0
+// CHECK: [[TMP3:%.*]] = load <2 x i64>, <2 x i64>* [[ARRAYIDX]], align 16
+// CHECK: [[TMP4:%.*]] = bitcast <2 x i64> [[TMP3]] to <16 x i8>
+// CHECK: [[VAL2:%.*]] = getelementptr inbounds %struct.poly64x2x2_t, %struct.poly64x2x2_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX3:%.*]] = getelementptr inbounds [2 x <2 x i64>], [2 x <2 x i64>]* [[VAL]]2, i64 0, i64 1
+// CHECK: [[TMP5:%.*]] = load <2 x i64>, <2 x i64>* [[ARRAYIDX]]3, align 16
+// CHECK: [[TMP6:%.*]] = bitcast <2 x i64> [[TMP5]] to <16 x i8>
+// CHECK: [[TMP7:%.*]] = bitcast <16 x i8> [[TMP4]] to <2 x i64>
+// CHECK: [[TMP8:%.*]] = bitcast <16 x i8> [[TMP6]] to <2 x i64>
+// CHECK: call void @llvm.aarch64.neon.st2.v2i64.p0i8(<2 x i64> [[TMP7]], <2 x i64> [[TMP8]], i8* [[TMP2]])
+// CHECK: ret void
void test_vst2q_p64(poly64_t * ptr, poly64x2x2_t val) {
- // CHECK-LABEL: test_vst2q_p64
return vst2q_p64(ptr, val);
- // CHECK: st2 {{{ *v[0-9]+.2d, v[0-9]+.2d *}}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define void @test_vst3_p64(i64* %ptr, [3 x <1 x i64>] %val.coerce) #0 {
+// CHECK: [[VAL:%.*]] = alloca %struct.poly64x1x3_t, align 8
+// CHECK: [[__S1:%.*]] = alloca %struct.poly64x1x3_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly64x1x3_t, %struct.poly64x1x3_t* [[VAL]], i32 0, i32 0
+// CHECK: store [3 x <1 x i64>] [[VAL]].coerce, [3 x <1 x i64>]* [[COERCE_DIVE]], align 8
+// CHECK: [[TMP0:%.*]] = bitcast %struct.poly64x1x3_t* [[__S1]] to i8*
+// CHECK: [[TMP1:%.*]] = bitcast %struct.poly64x1x3_t* [[VAL]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP0]], i8* [[TMP1]], i64 24, i32 8, i1 false)
+// CHECK: [[TMP2:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[VAL1:%.*]] = getelementptr inbounds %struct.poly64x1x3_t, %struct.poly64x1x3_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX:%.*]] = getelementptr inbounds [3 x <1 x i64>], [3 x <1 x i64>]* [[VAL]]1, i64 0, i64 0
+// CHECK: [[TMP3:%.*]] = load <1 x i64>, <1 x i64>* [[ARRAYIDX]], align 8
+// CHECK: [[TMP4:%.*]] = bitcast <1 x i64> [[TMP3]] to <8 x i8>
+// CHECK: [[VAL2:%.*]] = getelementptr inbounds %struct.poly64x1x3_t, %struct.poly64x1x3_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX3:%.*]] = getelementptr inbounds [3 x <1 x i64>], [3 x <1 x i64>]* [[VAL]]2, i64 0, i64 1
+// CHECK: [[TMP5:%.*]] = load <1 x i64>, <1 x i64>* [[ARRAYIDX]]3, align 8
+// CHECK: [[TMP6:%.*]] = bitcast <1 x i64> [[TMP5]] to <8 x i8>
+// CHECK: [[VAL4:%.*]] = getelementptr inbounds %struct.poly64x1x3_t, %struct.poly64x1x3_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX5:%.*]] = getelementptr inbounds [3 x <1 x i64>], [3 x <1 x i64>]* [[VAL]]4, i64 0, i64 2
+// CHECK: [[TMP7:%.*]] = load <1 x i64>, <1 x i64>* [[ARRAYIDX]]5, align 8
+// CHECK: [[TMP8:%.*]] = bitcast <1 x i64> [[TMP7]] to <8 x i8>
+// CHECK: [[TMP9:%.*]] = bitcast <8 x i8> [[TMP4]] to <1 x i64>
+// CHECK: [[TMP10:%.*]] = bitcast <8 x i8> [[TMP6]] to <1 x i64>
+// CHECK: [[TMP11:%.*]] = bitcast <8 x i8> [[TMP8]] to <1 x i64>
+// CHECK: call void @llvm.aarch64.neon.st3.v1i64.p0i8(<1 x i64> [[TMP9]], <1 x i64> [[TMP1]]0, <1 x i64> [[TMP1]]1, i8* [[TMP2]])
+// CHECK: ret void
void test_vst3_p64(poly64_t * ptr, poly64x1x3_t val) {
- // CHECK-LABEL: test_vst3_p64
return vst3_p64(ptr, val);
- // CHECK: st1 {{{ *v[0-9]+.1d, v[0-9]+.1d, v[0-9]+.1d *}}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define void @test_vst3q_p64(i64* %ptr, [3 x <2 x i64>] %val.coerce) #0 {
+// CHECK: [[VAL:%.*]] = alloca %struct.poly64x2x3_t, align 16
+// CHECK: [[__S1:%.*]] = alloca %struct.poly64x2x3_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly64x2x3_t, %struct.poly64x2x3_t* [[VAL]], i32 0, i32 0
+// CHECK: store [3 x <2 x i64>] [[VAL]].coerce, [3 x <2 x i64>]* [[COERCE_DIVE]], align 16
+// CHECK: [[TMP0:%.*]] = bitcast %struct.poly64x2x3_t* [[__S1]] to i8*
+// CHECK: [[TMP1:%.*]] = bitcast %struct.poly64x2x3_t* [[VAL]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP0]], i8* [[TMP1]], i64 48, i32 16, i1 false)
+// CHECK: [[TMP2:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[VAL1:%.*]] = getelementptr inbounds %struct.poly64x2x3_t, %struct.poly64x2x3_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX:%.*]] = getelementptr inbounds [3 x <2 x i64>], [3 x <2 x i64>]* [[VAL]]1, i64 0, i64 0
+// CHECK: [[TMP3:%.*]] = load <2 x i64>, <2 x i64>* [[ARRAYIDX]], align 16
+// CHECK: [[TMP4:%.*]] = bitcast <2 x i64> [[TMP3]] to <16 x i8>
+// CHECK: [[VAL2:%.*]] = getelementptr inbounds %struct.poly64x2x3_t, %struct.poly64x2x3_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX3:%.*]] = getelementptr inbounds [3 x <2 x i64>], [3 x <2 x i64>]* [[VAL]]2, i64 0, i64 1
+// CHECK: [[TMP5:%.*]] = load <2 x i64>, <2 x i64>* [[ARRAYIDX]]3, align 16
+// CHECK: [[TMP6:%.*]] = bitcast <2 x i64> [[TMP5]] to <16 x i8>
+// CHECK: [[VAL4:%.*]] = getelementptr inbounds %struct.poly64x2x3_t, %struct.poly64x2x3_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX5:%.*]] = getelementptr inbounds [3 x <2 x i64>], [3 x <2 x i64>]* [[VAL]]4, i64 0, i64 2
+// CHECK: [[TMP7:%.*]] = load <2 x i64>, <2 x i64>* [[ARRAYIDX]]5, align 16
+// CHECK: [[TMP8:%.*]] = bitcast <2 x i64> [[TMP7]] to <16 x i8>
+// CHECK: [[TMP9:%.*]] = bitcast <16 x i8> [[TMP4]] to <2 x i64>
+// CHECK: [[TMP10:%.*]] = bitcast <16 x i8> [[TMP6]] to <2 x i64>
+// CHECK: [[TMP11:%.*]] = bitcast <16 x i8> [[TMP8]] to <2 x i64>
+// CHECK: call void @llvm.aarch64.neon.st3.v2i64.p0i8(<2 x i64> [[TMP9]], <2 x i64> [[TMP1]]0, <2 x i64> [[TMP1]]1, i8* [[TMP2]])
+// CHECK: ret void
void test_vst3q_p64(poly64_t * ptr, poly64x2x3_t val) {
- // CHECK-LABEL: test_vst3q_p64
return vst3q_p64(ptr, val);
- // CHECK: st3 {{{ *v[0-9]+.2d, v[0-9]+.2d, v[0-9]+.2d *}}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define void @test_vst4_p64(i64* %ptr, [4 x <1 x i64>] %val.coerce) #0 {
+// CHECK: [[VAL:%.*]] = alloca %struct.poly64x1x4_t, align 8
+// CHECK: [[__S1:%.*]] = alloca %struct.poly64x1x4_t, align 8
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly64x1x4_t, %struct.poly64x1x4_t* [[VAL]], i32 0, i32 0
+// CHECK: store [4 x <1 x i64>] [[VAL]].coerce, [4 x <1 x i64>]* [[COERCE_DIVE]], align 8
+// CHECK: [[TMP0:%.*]] = bitcast %struct.poly64x1x4_t* [[__S1]] to i8*
+// CHECK: [[TMP1:%.*]] = bitcast %struct.poly64x1x4_t* [[VAL]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP0]], i8* [[TMP1]], i64 32, i32 8, i1 false)
+// CHECK: [[TMP2:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[VAL1:%.*]] = getelementptr inbounds %struct.poly64x1x4_t, %struct.poly64x1x4_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX:%.*]] = getelementptr inbounds [4 x <1 x i64>], [4 x <1 x i64>]* [[VAL]]1, i64 0, i64 0
+// CHECK: [[TMP3:%.*]] = load <1 x i64>, <1 x i64>* [[ARRAYIDX]], align 8
+// CHECK: [[TMP4:%.*]] = bitcast <1 x i64> [[TMP3]] to <8 x i8>
+// CHECK: [[VAL2:%.*]] = getelementptr inbounds %struct.poly64x1x4_t, %struct.poly64x1x4_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX3:%.*]] = getelementptr inbounds [4 x <1 x i64>], [4 x <1 x i64>]* [[VAL]]2, i64 0, i64 1
+// CHECK: [[TMP5:%.*]] = load <1 x i64>, <1 x i64>* [[ARRAYIDX]]3, align 8
+// CHECK: [[TMP6:%.*]] = bitcast <1 x i64> [[TMP5]] to <8 x i8>
+// CHECK: [[VAL4:%.*]] = getelementptr inbounds %struct.poly64x1x4_t, %struct.poly64x1x4_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX5:%.*]] = getelementptr inbounds [4 x <1 x i64>], [4 x <1 x i64>]* [[VAL]]4, i64 0, i64 2
+// CHECK: [[TMP7:%.*]] = load <1 x i64>, <1 x i64>* [[ARRAYIDX]]5, align 8
+// CHECK: [[TMP8:%.*]] = bitcast <1 x i64> [[TMP7]] to <8 x i8>
+// CHECK: [[VAL6:%.*]] = getelementptr inbounds %struct.poly64x1x4_t, %struct.poly64x1x4_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX7:%.*]] = getelementptr inbounds [4 x <1 x i64>], [4 x <1 x i64>]* [[VAL]]6, i64 0, i64 3
+// CHECK: [[TMP9:%.*]] = load <1 x i64>, <1 x i64>* [[ARRAYIDX]]7, align 8
+// CHECK: [[TMP10:%.*]] = bitcast <1 x i64> [[TMP9]] to <8 x i8>
+// CHECK: [[TMP11:%.*]] = bitcast <8 x i8> [[TMP4]] to <1 x i64>
+// CHECK: [[TMP12:%.*]] = bitcast <8 x i8> [[TMP6]] to <1 x i64>
+// CHECK: [[TMP13:%.*]] = bitcast <8 x i8> [[TMP8]] to <1 x i64>
+// CHECK: [[TMP14:%.*]] = bitcast <8 x i8> [[TMP1]]0 to <1 x i64>
+// CHECK: call void @llvm.aarch64.neon.st4.v1i64.p0i8(<1 x i64> [[TMP1]]1, <1 x i64> [[TMP1]]2, <1 x i64> [[TMP1]]3, <1 x i64> [[TMP1]]4, i8* [[TMP2]])
+// CHECK: ret void
void test_vst4_p64(poly64_t * ptr, poly64x1x4_t val) {
- // CHECK-LABEL: test_vst4_p64
return vst4_p64(ptr, val);
- // CHECK: st1 {{{ *v[0-9]+.1d, v[0-9]+.1d, v[0-9]+.1d, v[0-9]+.1d *}}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define void @test_vst4q_p64(i64* %ptr, [4 x <2 x i64>] %val.coerce) #0 {
+// CHECK: [[VAL:%.*]] = alloca %struct.poly64x2x4_t, align 16
+// CHECK: [[__S1:%.*]] = alloca %struct.poly64x2x4_t, align 16
+// CHECK: [[COERCE_DIVE:%.*]] = getelementptr inbounds %struct.poly64x2x4_t, %struct.poly64x2x4_t* [[VAL]], i32 0, i32 0
+// CHECK: store [4 x <2 x i64>] [[VAL]].coerce, [4 x <2 x i64>]* [[COERCE_DIVE]], align 16
+// CHECK: [[TMP0:%.*]] = bitcast %struct.poly64x2x4_t* [[__S1]] to i8*
+// CHECK: [[TMP1:%.*]] = bitcast %struct.poly64x2x4_t* [[VAL]] to i8*
+// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64(i8* [[TMP0]], i8* [[TMP1]], i64 64, i32 16, i1 false)
+// CHECK: [[TMP2:%.*]] = bitcast i64* %ptr to i8*
+// CHECK: [[VAL1:%.*]] = getelementptr inbounds %struct.poly64x2x4_t, %struct.poly64x2x4_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX:%.*]] = getelementptr inbounds [4 x <2 x i64>], [4 x <2 x i64>]* [[VAL]]1, i64 0, i64 0
+// CHECK: [[TMP3:%.*]] = load <2 x i64>, <2 x i64>* [[ARRAYIDX]], align 16
+// CHECK: [[TMP4:%.*]] = bitcast <2 x i64> [[TMP3]] to <16 x i8>
+// CHECK: [[VAL2:%.*]] = getelementptr inbounds %struct.poly64x2x4_t, %struct.poly64x2x4_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX3:%.*]] = getelementptr inbounds [4 x <2 x i64>], [4 x <2 x i64>]* [[VAL]]2, i64 0, i64 1
+// CHECK: [[TMP5:%.*]] = load <2 x i64>, <2 x i64>* [[ARRAYIDX]]3, align 16
+// CHECK: [[TMP6:%.*]] = bitcast <2 x i64> [[TMP5]] to <16 x i8>
+// CHECK: [[VAL4:%.*]] = getelementptr inbounds %struct.poly64x2x4_t, %struct.poly64x2x4_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX5:%.*]] = getelementptr inbounds [4 x <2 x i64>], [4 x <2 x i64>]* [[VAL]]4, i64 0, i64 2
+// CHECK: [[TMP7:%.*]] = load <2 x i64>, <2 x i64>* [[ARRAYIDX]]5, align 16
+// CHECK: [[TMP8:%.*]] = bitcast <2 x i64> [[TMP7]] to <16 x i8>
+// CHECK: [[VAL6:%.*]] = getelementptr inbounds %struct.poly64x2x4_t, %struct.poly64x2x4_t* [[__S1]], i32 0, i32 0
+// CHECK: [[ARRAYIDX7:%.*]] = getelementptr inbounds [4 x <2 x i64>], [4 x <2 x i64>]* [[VAL]]6, i64 0, i64 3
+// CHECK: [[TMP9:%.*]] = load <2 x i64>, <2 x i64>* [[ARRAYIDX]]7, align 16
+// CHECK: [[TMP10:%.*]] = bitcast <2 x i64> [[TMP9]] to <16 x i8>
+// CHECK: [[TMP11:%.*]] = bitcast <16 x i8> [[TMP4]] to <2 x i64>
+// CHECK: [[TMP12:%.*]] = bitcast <16 x i8> [[TMP6]] to <2 x i64>
+// CHECK: [[TMP13:%.*]] = bitcast <16 x i8> [[TMP8]] to <2 x i64>
+// CHECK: [[TMP14:%.*]] = bitcast <16 x i8> [[TMP1]]0 to <2 x i64>
+// CHECK: call void @llvm.aarch64.neon.st4.v2i64.p0i8(<2 x i64> [[TMP1]]1, <2 x i64> [[TMP1]]2, <2 x i64> [[TMP1]]3, <2 x i64> [[TMP1]]4, i8* [[TMP2]])
+// CHECK: ret void
void test_vst4q_p64(poly64_t * ptr, poly64x2x4_t val) {
- // CHECK-LABEL: test_vst4q_p64
return vst4q_p64(ptr, val);
- // CHECK: st4 {{{ *v[0-9]+.2d, v[0-9]+.2d, v[0-9]+.2d, v[0-9]+.2d *}}}, [{{x[0-9]+|sp}}]
}
+// CHECK-LABEL: define <1 x i64> @test_vext_p64(<1 x i64> %a, <1 x i64> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <1 x i64> %b to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP1]] to <1 x i64>
+// CHECK: [[VEXT:%.*]] = shufflevector <1 x i64> [[TMP2]], <1 x i64> [[TMP3]], <1 x i32> zeroinitializer
+// CHECK: ret <1 x i64> [[VEXT]]
poly64x1_t test_vext_p64(poly64x1_t a, poly64x1_t b) {
- // CHECK-LABEL: test_vext_p64
return vext_u64(a, b, 0);
}
+// CHECK-LABEL: define <2 x i64> @test_vextq_p64(<2 x i64> %a, <2 x i64> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <2 x i64> %b to <16 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP1]] to <2 x i64>
+// CHECK: [[VEXT:%.*]] = shufflevector <2 x i64> [[TMP2]], <2 x i64> [[TMP3]], <2 x i32> <i32 1, i32 2>
+// CHECK: ret <2 x i64> [[VEXT]]
poly64x2_t test_vextq_p64(poly64x2_t a, poly64x2_t b) {
- // CHECK-LABEL: test_vextq_p64
return vextq_p64(a, b, 1);
- // CHECK: ext {{v[0-9]+}}.16b, {{v[0-9]+}}.16b, {{v[0-9]+}}.16b, {{#0x8|#8}}
}
+// CHECK-LABEL: define <2 x i64> @test_vzip1q_p64(<2 x i64> %a, <2 x i64> %b) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 0, i32 2>
+// CHECK: ret <2 x i64> [[SHUFFLE_I]]
poly64x2_t test_vzip1q_p64(poly64x2_t a, poly64x2_t b) {
- // CHECK-LABEL: test_vzip1q_p64
return vzip1q_p64(a, b);
- // CHECK-ARM64: zip1 {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
}
+// CHECK-LABEL: define <2 x i64> @test_vzip2q_p64(<2 x i64> %a, <2 x i64> %b) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 1, i32 3>
+// CHECK: ret <2 x i64> [[SHUFFLE_I]]
poly64x2_t test_vzip2q_p64(poly64x2_t a, poly64x2_t b) {
- // CHECK-LABEL: test_vzip2q_p64
return vzip2q_u64(a, b);
- // CHECK-ARM64: zip2 {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
}
+// CHECK-LABEL: define <2 x i64> @test_vuzp1q_p64(<2 x i64> %a, <2 x i64> %b) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 0, i32 2>
+// CHECK: ret <2 x i64> [[SHUFFLE_I]]
poly64x2_t test_vuzp1q_p64(poly64x2_t a, poly64x2_t b) {
- // CHECK-LABEL: test_vuzp1q_p64
return vuzp1q_p64(a, b);
- // CHECK-ARM64: zip1 {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
}
+// CHECK-LABEL: define <2 x i64> @test_vuzp2q_p64(<2 x i64> %a, <2 x i64> %b) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 1, i32 3>
+// CHECK: ret <2 x i64> [[SHUFFLE_I]]
poly64x2_t test_vuzp2q_p64(poly64x2_t a, poly64x2_t b) {
- // CHECK-LABEL: test_vuzp2q_p64
return vuzp2q_u64(a, b);
- // CHECK-ARM64: zip2 {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
}
+// CHECK-LABEL: define <2 x i64> @test_vtrn1q_p64(<2 x i64> %a, <2 x i64> %b) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 0, i32 2>
+// CHECK: ret <2 x i64> [[SHUFFLE_I]]
poly64x2_t test_vtrn1q_p64(poly64x2_t a, poly64x2_t b) {
- // CHECK-LABEL: test_vtrn1q_p64
return vtrn1q_p64(a, b);
- // CHECK-ARM64: zip1 {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
}
+// CHECK-LABEL: define <2 x i64> @test_vtrn2q_p64(<2 x i64> %a, <2 x i64> %b) #0 {
+// CHECK: [[SHUFFLE_I:%.*]] = shufflevector <2 x i64> %a, <2 x i64> %b, <2 x i32> <i32 1, i32 3>
+// CHECK: ret <2 x i64> [[SHUFFLE_I]]
poly64x2_t test_vtrn2q_p64(poly64x2_t a, poly64x2_t b) {
- // CHECK-LABEL: test_vtrn2q_p64
return vtrn2q_u64(a, b);
- // CHECK-ARM64: zip2 {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, {{v[0-9]+}}.2d
}
+// CHECK-LABEL: define <1 x i64> @test_vsri_n_p64(<1 x i64> %a, <1 x i64> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <1 x i64> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <1 x i64> %b to <8 x i8>
+// CHECK: [[VSRI_N:%.*]] = bitcast <8 x i8> [[TMP0]] to <1 x i64>
+// CHECK: [[VSRI_N1:%.*]] = bitcast <8 x i8> [[TMP1]] to <1 x i64>
+// CHECK: [[VSRI_N2:%.*]] = call <1 x i64> @llvm.aarch64.neon.vsri.v1i64(<1 x i64> [[VSRI_N]], <1 x i64> [[VSRI_N]]1, i32 33)
+// CHECK: ret <1 x i64> [[VSRI_N]]2
poly64x1_t test_vsri_n_p64(poly64x1_t a, poly64x1_t b) {
- // CHECK-LABEL: test_vsri_n_p64
return vsri_n_p64(a, b, 33);
- // CHECK: sri {{d[0-9]+}}, {{d[0-9]+}}, #33
}
+// CHECK-LABEL: define <2 x i64> @test_vsriq_n_p64(<2 x i64> %a, <2 x i64> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <2 x i64> %b to <16 x i8>
+// CHECK: [[VSRI_N:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64>
+// CHECK: [[VSRI_N1:%.*]] = bitcast <16 x i8> [[TMP1]] to <2 x i64>
+// CHECK: [[VSRI_N2:%.*]] = call <2 x i64> @llvm.aarch64.neon.vsri.v2i64(<2 x i64> [[VSRI_N]], <2 x i64> [[VSRI_N]]1, i32 64)
+// CHECK: ret <2 x i64> [[VSRI_N]]2
poly64x2_t test_vsriq_n_p64(poly64x2_t a, poly64x2_t b) {
- // CHECK-LABEL: test_vsriq_n_p64
return vsriq_n_p64(a, b, 64);
- // CHECK: sri {{v[0-9]+}}.2d, {{v[0-9]+}}.2d, #64
}
Modified: cfe/trunk/test/CodeGen/arm-bitfield-alignment.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm-bitfield-alignment.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/arm-bitfield-alignment.c (original)
+++ cfe/trunk/test/CodeGen/arm-bitfield-alignment.c Wed Mar 9 12:54:42 2016
@@ -1,5 +1,5 @@
-// RUN: %clang_cc1 -triple arm-none-eabi -ffreestanding -emit-llvm -o - -O3 %s | FileCheck %s
-// RUN: %clang_cc1 -triple aarch64 -ffreestanding -emit-llvm -o - -O3 %s | FileCheck %s
+// RUN: %clang_cc1 -triple arm-none-eabi -ffreestanding -emit-llvm -o - %s | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64 -ffreestanding -emit-llvm -o - %s | FileCheck %s
extern struct T {
int b0 : 8;
Modified: cfe/trunk/test/CodeGen/arm-crc32.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm-crc32.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/arm-crc32.c (original)
+++ cfe/trunk/test/CodeGen/arm-crc32.c Wed Mar 9 12:54:42 2016
@@ -1,6 +1,5 @@
-// REQUIRES: arm-registered-target
// RUN: %clang_cc1 -triple armv8-none-linux-gnueabi \
-// RUN: -O3 -S -emit-llvm -o - %s | FileCheck %s
+// RUN: -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s
int crc32b(int a, char b)
{
@@ -48,7 +47,7 @@ int crc32d(int a, long long b)
// CHECK: [[T0:%[0-9]+]] = trunc i64 %b to i32
// CHECK: [[T1:%[0-9]+]] = lshr i64 %b, 32
// CHECK: [[T2:%[0-9]+]] = trunc i64 [[T1]] to i32
-// CHECK: [[T3:%[0-9]+]] = tail call i32 @llvm.arm.crc32w(i32 %a, i32 [[T0]])
+// CHECK: [[T3:%[0-9]+]] = call i32 @llvm.arm.crc32w(i32 %a, i32 [[T0]])
// CHECK: call i32 @llvm.arm.crc32w(i32 [[T3]], i32 [[T2]])
}
@@ -58,6 +57,6 @@ int crc32cd(int a, long long b)
// CHECK: [[T0:%[0-9]+]] = trunc i64 %b to i32
// CHECK: [[T1:%[0-9]+]] = lshr i64 %b, 32
// CHECK: [[T2:%[0-9]+]] = trunc i64 [[T1]] to i32
-// CHECK: [[T3:%[0-9]+]] = tail call i32 @llvm.arm.crc32cw(i32 %a, i32 [[T0]])
+// CHECK: [[T3:%[0-9]+]] = call i32 @llvm.arm.crc32cw(i32 %a, i32 [[T0]])
// CHECK: call i32 @llvm.arm.crc32cw(i32 [[T3]], i32 [[T2]])
}
Modified: cfe/trunk/test/CodeGen/arm-neon-directed-rounding.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm-neon-directed-rounding.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/arm-neon-directed-rounding.c (original)
+++ cfe/trunk/test/CodeGen/arm-neon-directed-rounding.c Wed Mar 9 12:54:42 2016
@@ -1,75 +1,135 @@
-// RUN: %clang_cc1 -triple thumbv8-linux-gnueabihf -target-cpu cortex-a57 -ffreestanding -O1 -emit-llvm %s -o - | FileCheck %s
+// RUN: %clang_cc1 -triple thumbv8-linux-gnueabihf -target-cpu cortex-a57 -ffreestanding -emit-llvm %s -o - | opt -S -mem2reg | FileCheck %s
#include <arm_neon.h>
+// CHECK-LABEL: define <2 x float> @test_vrnda_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[VRNDA_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VRNDA_V1_I:%.*]] = call <2 x float> @llvm.arm.neon.vrinta.v2f32(<2 x float> [[VRNDA_V_I]]) #2
+// CHECK: [[VRNDA_V2_I:%.*]] = bitcast <2 x float> [[VRNDA_V1_I]] to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[VRNDA_V2_I]] to <2 x float>
+// CHECK: ret <2 x float> [[TMP1]]
float32x2_t test_vrnda_f32(float32x2_t a) {
- // CHECK-LABEL: test_vrnda_f32
- // CHECK: call <2 x float> @llvm.arm.neon.vrinta.v2f32(<2 x float> %a)
return vrnda_f32(a);
}
+// CHECK-LABEL: define <4 x float> @test_vrndaq_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[VRNDAQ_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VRNDAQ_V1_I:%.*]] = call <4 x float> @llvm.arm.neon.vrinta.v4f32(<4 x float> [[VRNDAQ_V_I]]) #2
+// CHECK: [[VRNDAQ_V2_I:%.*]] = bitcast <4 x float> [[VRNDAQ_V1_I]] to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[VRNDAQ_V2_I]] to <4 x float>
+// CHECK: ret <4 x float> [[TMP1]]
float32x4_t test_vrndaq_f32(float32x4_t a) {
- // CHECK-LABEL: test_vrndaq_f32
- // CHECK: call <4 x float> @llvm.arm.neon.vrinta.v4f32(<4 x float> %a)
return vrndaq_f32(a);
}
+// CHECK-LABEL: define <2 x float> @test_vrndm_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[VRNDM_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VRNDM_V1_I:%.*]] = call <2 x float> @llvm.arm.neon.vrintm.v2f32(<2 x float> [[VRNDM_V_I]]) #2
+// CHECK: [[VRNDM_V2_I:%.*]] = bitcast <2 x float> [[VRNDM_V1_I]] to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[VRNDM_V2_I]] to <2 x float>
+// CHECK: ret <2 x float> [[TMP1]]
float32x2_t test_vrndm_f32(float32x2_t a) {
- // CHECK-LABEL: test_vrndm_f32
- // CHECK: call <2 x float> @llvm.arm.neon.vrintm.v2f32(<2 x float> %a)
return vrndm_f32(a);
}
+// CHECK-LABEL: define <4 x float> @test_vrndmq_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[VRNDMQ_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VRNDMQ_V1_I:%.*]] = call <4 x float> @llvm.arm.neon.vrintm.v4f32(<4 x float> [[VRNDMQ_V_I]]) #2
+// CHECK: [[VRNDMQ_V2_I:%.*]] = bitcast <4 x float> [[VRNDMQ_V1_I]] to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[VRNDMQ_V2_I]] to <4 x float>
+// CHECK: ret <4 x float> [[TMP1]]
float32x4_t test_vrndmq_f32(float32x4_t a) {
- // CHECK-LABEL: test_vrndmq_f32
- // CHECK: call <4 x float> @llvm.arm.neon.vrintm.v4f32(<4 x float> %a)
return vrndmq_f32(a);
}
+// CHECK-LABEL: define <2 x float> @test_vrndn_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[VRNDN_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VRNDN_V1_I:%.*]] = call <2 x float> @llvm.arm.neon.vrintn.v2f32(<2 x float> [[VRNDN_V_I]]) #2
+// CHECK: [[VRNDN_V2_I:%.*]] = bitcast <2 x float> [[VRNDN_V1_I]] to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[VRNDN_V2_I]] to <2 x float>
+// CHECK: ret <2 x float> [[TMP1]]
float32x2_t test_vrndn_f32(float32x2_t a) {
- // CHECK-LABEL: test_vrndn_f32
- // CHECK: call <2 x float> @llvm.arm.neon.vrintn.v2f32(<2 x float> %a)
return vrndn_f32(a);
}
+// CHECK-LABEL: define <4 x float> @test_vrndnq_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[VRNDNQ_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VRNDNQ_V1_I:%.*]] = call <4 x float> @llvm.arm.neon.vrintn.v4f32(<4 x float> [[VRNDNQ_V_I]]) #2
+// CHECK: [[VRNDNQ_V2_I:%.*]] = bitcast <4 x float> [[VRNDNQ_V1_I]] to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[VRNDNQ_V2_I]] to <4 x float>
+// CHECK: ret <4 x float> [[TMP1]]
float32x4_t test_vrndnq_f32(float32x4_t a) {
- // CHECK-LABEL: test_vrndnq_f32
- // CHECK: call <4 x float> @llvm.arm.neon.vrintn.v4f32(<4 x float> %a)
return vrndnq_f32(a);
}
+// CHECK-LABEL: define <2 x float> @test_vrndp_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[VRNDP_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VRNDP_V1_I:%.*]] = call <2 x float> @llvm.arm.neon.vrintp.v2f32(<2 x float> [[VRNDP_V_I]]) #2
+// CHECK: [[VRNDP_V2_I:%.*]] = bitcast <2 x float> [[VRNDP_V1_I]] to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[VRNDP_V2_I]] to <2 x float>
+// CHECK: ret <2 x float> [[TMP1]]
float32x2_t test_vrndp_f32(float32x2_t a) {
- // CHECK-LABEL: test_vrndp_f32
- // CHECK: call <2 x float> @llvm.arm.neon.vrintp.v2f32(<2 x float> %a)
return vrndp_f32(a);
}
+// CHECK-LABEL: define <4 x float> @test_vrndpq_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[VRNDPQ_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VRNDPQ_V1_I:%.*]] = call <4 x float> @llvm.arm.neon.vrintp.v4f32(<4 x float> [[VRNDPQ_V_I]]) #2
+// CHECK: [[VRNDPQ_V2_I:%.*]] = bitcast <4 x float> [[VRNDPQ_V1_I]] to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[VRNDPQ_V2_I]] to <4 x float>
+// CHECK: ret <4 x float> [[TMP1]]
float32x4_t test_vrndpq_f32(float32x4_t a) {
- // CHECK-LABEL: test_vrndpq_f32
- // CHECK: call <4 x float> @llvm.arm.neon.vrintp.v4f32(<4 x float> %a)
return vrndpq_f32(a);
}
+// CHECK-LABEL: define <2 x float> @test_vrndx_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[VRNDX_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VRNDX_V1_I:%.*]] = call <2 x float> @llvm.arm.neon.vrintx.v2f32(<2 x float> [[VRNDX_V_I]]) #2
+// CHECK: [[VRNDX_V2_I:%.*]] = bitcast <2 x float> [[VRNDX_V1_I]] to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[VRNDX_V2_I]] to <2 x float>
+// CHECK: ret <2 x float> [[TMP1]]
float32x2_t test_vrndx_f32(float32x2_t a) {
- // CHECK-LABEL: test_vrndx_f32
- // CHECK: call <2 x float> @llvm.arm.neon.vrintx.v2f32(<2 x float> %a)
return vrndx_f32(a);
}
+// CHECK-LABEL: define <4 x float> @test_vrndxq_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[VRNDXQ_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VRNDXQ_V1_I:%.*]] = call <4 x float> @llvm.arm.neon.vrintx.v4f32(<4 x float> [[VRNDXQ_V_I]]) #2
+// CHECK: [[VRNDXQ_V2_I:%.*]] = bitcast <4 x float> [[VRNDXQ_V1_I]] to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[VRNDXQ_V2_I]] to <4 x float>
+// CHECK: ret <4 x float> [[TMP1]]
float32x4_t test_vrndxq_f32(float32x4_t a) {
- // CHECK-LABEL: test_vrndxq_f32
- // CHECK: call <4 x float> @llvm.arm.neon.vrintx.v4f32(<4 x float> %a)
return vrndxq_f32(a);
}
+// CHECK-LABEL: define <2 x float> @test_vrnd_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[VRND_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VRND_V1_I:%.*]] = call <2 x float> @llvm.arm.neon.vrintz.v2f32(<2 x float> [[VRND_V_I]]) #2
+// CHECK: [[VRND_V2_I:%.*]] = bitcast <2 x float> [[VRND_V1_I]] to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <8 x i8> [[VRND_V2_I]] to <2 x float>
+// CHECK: ret <2 x float> [[TMP1]]
float32x2_t test_vrnd_f32(float32x2_t a) {
- // CHECK-LABEL: test_vrnd_f32
- // CHECK: call <2 x float> @llvm.arm.neon.vrintz.v2f32(<2 x float> %a)
return vrnd_f32(a);
}
+// CHECK-LABEL: define <4 x float> @test_vrndq_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[VRNDQ_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VRNDQ_V1_I:%.*]] = call <4 x float> @llvm.arm.neon.vrintz.v4f32(<4 x float> [[VRNDQ_V_I]]) #2
+// CHECK: [[VRNDQ_V2_I:%.*]] = bitcast <4 x float> [[VRNDQ_V1_I]] to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[VRNDQ_V2_I]] to <4 x float>
+// CHECK: ret <4 x float> [[TMP1]]
float32x4_t test_vrndq_f32(float32x4_t a) {
- // CHECK-LABEL: test_vrndq_f32
- // CHECK: call <4 x float> @llvm.arm.neon.vrintz.v4f32(<4 x float> %a)
return vrndq_f32(a);
}
Modified: cfe/trunk/test/CodeGen/arm-neon-fma.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm-neon-fma.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/arm-neon-fma.c (original)
+++ cfe/trunk/test/CodeGen/arm-neon-fma.c Wed Mar 9 12:54:42 2016
@@ -1,19 +1,34 @@
-// REQUIRES: arm-registered-target
// RUN: %clang_cc1 -triple thumbv7-none-linux-gnueabihf \
// RUN: -target-abi aapcs \
// RUN: -target-cpu cortex-a8 \
// RUN: -mfloat-abi hard \
// RUN: -ffreestanding \
-// RUN: -O3 -S -emit-llvm -o - %s | FileCheck %s
+// RUN: -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s
#include <arm_neon.h>
+// CHECK-LABEL: define <2 x float> @test_fma_order(<2 x float> %accum, <2 x float> %lhs, <2 x float> %rhs) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %accum to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <2 x float> %lhs to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <2 x float> %rhs to <8 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[TMP4:%.*]] = bitcast <8 x i8> [[TMP1]] to <2 x float>
+// CHECK: [[TMP5:%.*]] = bitcast <8 x i8> [[TMP2]] to <2 x float>
+// CHECK: [[TMP6:%.*]] = call <2 x float> @llvm.fma.v2f32(<2 x float> [[TMP4]], <2 x float> [[TMP5]], <2 x float> [[TMP3]]) #2
+// CHECK: ret <2 x float> [[TMP6]]
float32x2_t test_fma_order(float32x2_t accum, float32x2_t lhs, float32x2_t rhs) {
return vfma_f32(accum, lhs, rhs);
-// CHECK: call <2 x float> @llvm.fma.v2f32(<2 x float> %lhs, <2 x float> %rhs, <2 x float> %accum)
}
+// CHECK-LABEL: define <4 x float> @test_fmaq_order(<4 x float> %accum, <4 x float> %lhs, <4 x float> %rhs) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %accum to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <4 x float> %lhs to <16 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <4 x float> %rhs to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[TMP4:%.*]] = bitcast <16 x i8> [[TMP1]] to <4 x float>
+// CHECK: [[TMP5:%.*]] = bitcast <16 x i8> [[TMP2]] to <4 x float>
+// CHECK: [[TMP6:%.*]] = call <4 x float> @llvm.fma.v4f32(<4 x float> [[TMP4]], <4 x float> [[TMP5]], <4 x float> [[TMP3]]) #2
+// CHECK: ret <4 x float> [[TMP6]]
float32x4_t test_fmaq_order(float32x4_t accum, float32x4_t lhs, float32x4_t rhs) {
return vfmaq_f32(accum, lhs, rhs);
-// CHECK: call <4 x float> @llvm.fma.v4f32(<4 x float> %lhs, <4 x float> %rhs, <4 x float> %accum)
}
Modified: cfe/trunk/test/CodeGen/arm-neon-numeric-maxmin.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm-neon-numeric-maxmin.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/arm-neon-numeric-maxmin.c (original)
+++ cfe/trunk/test/CodeGen/arm-neon-numeric-maxmin.c Wed Mar 9 12:54:42 2016
@@ -1,27 +1,55 @@
-// RUN: %clang_cc1 -triple thumbv8-linux-gnueabihf -target-cpu cortex-a57 -ffreestanding -O1 -emit-llvm %s -o - | FileCheck %s
+// RUN: %clang_cc1 -triple thumbv8-linux-gnueabihf -target-cpu cortex-a57 -ffreestanding -emit-llvm %s -o - | opt -S -mem2reg | FileCheck %s
#include <arm_neon.h>
+// CHECK-LABEL: define <2 x float> @test_vmaxnm_f32(<2 x float> %a, <2 x float> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <2 x float> %b to <8 x i8>
+// CHECK: [[VMAXNM_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VMAXNM_V1_I:%.*]] = bitcast <8 x i8> [[TMP1]] to <2 x float>
+// CHECK: [[VMAXNM_V2_I:%.*]] = call <2 x float> @llvm.arm.neon.vmaxnm.v2f32(<2 x float> [[VMAXNM_V_I]], <2 x float> [[VMAXNM_V1_I]]) #2
+// CHECK: [[VMAXNM_V3_I:%.*]] = bitcast <2 x float> [[VMAXNM_V2_I]] to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <8 x i8> [[VMAXNM_V3_I]] to <2 x float>
+// CHECK: ret <2 x float> [[TMP2]]
float32x2_t test_vmaxnm_f32(float32x2_t a, float32x2_t b) {
- // CHECK-LABEL: test_vmaxnm_f32
- // CHECK: call <2 x float> @llvm.arm.neon.vmaxnm.v2f32(<2 x float> %a, <2 x float> %b)
return vmaxnm_f32(a, b);
}
+// CHECK-LABEL: define <4 x float> @test_vmaxnmq_f32(<4 x float> %a, <4 x float> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <4 x float> %b to <16 x i8>
+// CHECK: [[VMAXNMQ_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VMAXNMQ_V1_I:%.*]] = bitcast <16 x i8> [[TMP1]] to <4 x float>
+// CHECK: [[VMAXNMQ_V2_I:%.*]] = call <4 x float> @llvm.arm.neon.vmaxnm.v4f32(<4 x float> [[VMAXNMQ_V_I]], <4 x float> [[VMAXNMQ_V1_I]]) #2
+// CHECK: [[VMAXNMQ_V3_I:%.*]] = bitcast <4 x float> [[VMAXNMQ_V2_I]] to <16 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[VMAXNMQ_V3_I]] to <4 x float>
+// CHECK: ret <4 x float> [[TMP2]]
float32x4_t test_vmaxnmq_f32(float32x4_t a, float32x4_t b) {
- // CHECK-LABEL: test_vmaxnmq_f32
- // CHECK: call <4 x float> @llvm.arm.neon.vmaxnm.v4f32(<4 x float> %a, <4 x float> %b)
return vmaxnmq_f32(a, b);
}
+// CHECK-LABEL: define <2 x float> @test_vminnm_f32(<2 x float> %a, <2 x float> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <2 x float> %b to <8 x i8>
+// CHECK: [[VMINNM_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VMINNM_V1_I:%.*]] = bitcast <8 x i8> [[TMP1]] to <2 x float>
+// CHECK: [[VMINNM_V2_I:%.*]] = call <2 x float> @llvm.arm.neon.vminnm.v2f32(<2 x float> [[VMINNM_V_I]], <2 x float> [[VMINNM_V1_I]]) #2
+// CHECK: [[VMINNM_V3_I:%.*]] = bitcast <2 x float> [[VMINNM_V2_I]] to <8 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <8 x i8> [[VMINNM_V3_I]] to <2 x float>
+// CHECK: ret <2 x float> [[TMP2]]
float32x2_t test_vminnm_f32(float32x2_t a, float32x2_t b) {
- // CHECK-LABEL: test_vminnm_f32
- // CHECK: call <2 x float> @llvm.arm.neon.vminnm.v2f32(<2 x float> %a, <2 x float> %b)
return vminnm_f32(a, b);
}
+// CHECK-LABEL: define <4 x float> @test_vminnmq_f32(<4 x float> %a, <4 x float> %b) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <4 x float> %b to <16 x i8>
+// CHECK: [[VMINNMQ_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VMINNMQ_V1_I:%.*]] = bitcast <16 x i8> [[TMP1]] to <4 x float>
+// CHECK: [[VMINNMQ_V2_I:%.*]] = call <4 x float> @llvm.arm.neon.vminnm.v4f32(<4 x float> [[VMINNMQ_V_I]], <4 x float> [[VMINNMQ_V1_I]]) #2
+// CHECK: [[VMINNMQ_V3_I:%.*]] = bitcast <4 x float> [[VMINNMQ_V2_I]] to <16 x i8>
+// CHECK: [[TMP2:%.*]] = bitcast <16 x i8> [[VMINNMQ_V3_I]] to <4 x float>
+// CHECK: ret <4 x float> [[TMP2]]
float32x4_t test_vminnmq_f32(float32x4_t a, float32x4_t b) {
- // CHECK-LABEL: test_vminnmq_f32
- // CHECK: call <4 x float> @llvm.arm.neon.vminnm.v4f32(<4 x float> %a, <4 x float> %b)
return vminnmq_f32(a, b);
}
Modified: cfe/trunk/test/CodeGen/arm-neon-shifts.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm-neon-shifts.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/arm-neon-shifts.c (original)
+++ cfe/trunk/test/CodeGen/arm-neon-shifts.c Wed Mar 9 12:54:42 2016
@@ -2,7 +2,7 @@
// RUN: %clang_cc1 -triple thumbv7-apple-darwin \
// RUN: -target-cpu cortex-a8 \
// RUN: -ffreestanding \
-// RUN: -emit-llvm -w -O1 -o - %s | FileCheck %s
+// RUN: -emit-llvm -w -o - %s | opt -S -mem2reg | FileCheck %s
#include <arm_neon.h>
@@ -27,19 +27,20 @@ uint8x8_t test_shift_vshr_umax(uint8x8_t
uint8x8_t test_shift_vsra(uint8x8_t a, uint8x8_t b) {
// CHECK-LABEL: test_shift_vsra
// CHECK: %[[SHR:.*]] = lshr <8 x i8> %b, <i8 5, i8 5, i8 5, i8 5, i8 5, i8 5, i8 5, i8 5>
- // CHECK: %{{.*}} = add <8 x i8> %[[SHR]], %a
+ // CHECK: %{{.*}} = add <8 x i8> %a, %[[SHR]]
return vsra_n_u8(a, b, 5);
}
int8x8_t test_shift_vsra_smax(int8x8_t a, int8x8_t b) {
// CHECK-LABEL: test_shift_vsra_smax
// CHECK: %[[SHR:.*]] = ashr <8 x i8> %b, <i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7>
- // CHECK: %{{.*}} = add <8 x i8> %[[SHR]], %a
+ // CHECK: %{{.*}} = add <8 x i8> %a, %[[SHR]]
return vsra_n_s8(a, b, 8);
}
uint8x8_t test_shift_vsra_umax(uint8x8_t a, uint8x8_t b) {
// CHECK-LABEL: test_shift_vsra_umax
- // CHECK: ret <8 x i8> %a
+ // CHECK: [[RES:%.*]] = add <8 x i8> %a, zeroinitializer
+ // CHECK: ret <8 x i8> [[RES]]
return vsra_n_u8(a, b, 8);
}
Modified: cfe/trunk/test/CodeGen/arm-neon-vcvtX.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm-neon-vcvtX.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/arm-neon-vcvtX.c (original)
+++ cfe/trunk/test/CodeGen/arm-neon-vcvtX.c Wed Mar 9 12:54:42 2016
@@ -1,99 +1,147 @@
-// RUN: %clang_cc1 -triple thumbv8-linux-gnueabihf -target-cpu cortex-a57 -ffreestanding -O1 -emit-llvm %s -o - | FileCheck %s
+// RUN: %clang_cc1 -triple thumbv8-linux-gnueabihf -target-cpu cortex-a57 -ffreestanding -emit-llvm %s -o - | opt -S -mem2reg | FileCheck %s
#include <arm_neon.h>
+// CHECK-LABEL: define <2 x i32> @test_vcvta_s32_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[VCVTA_S32_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VCVTA_S32_V1_I:%.*]] = call <2 x i32> @llvm.arm.neon.vcvtas.v2i32.v2f32(<2 x float> [[VCVTA_S32_V_I]]) #2
+// CHECK: ret <2 x i32> [[VCVTA_S32_V1_I]]
int32x2_t test_vcvta_s32_f32(float32x2_t a) {
- // CHECK-LABEL: test_vcvta_s32_f32
- // CHECK-LABEL: call <2 x i32> @llvm.arm.neon.vcvtas.v2i32.v2f32(<2 x float> %a)
return vcvta_s32_f32(a);
}
+// CHECK-LABEL: define <2 x i32> @test_vcvta_u32_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[VCVTA_U32_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VCVTA_U32_V1_I:%.*]] = call <2 x i32> @llvm.arm.neon.vcvtau.v2i32.v2f32(<2 x float> [[VCVTA_U32_V_I]]) #2
+// CHECK: ret <2 x i32> [[VCVTA_U32_V1_I]]
uint32x2_t test_vcvta_u32_f32(float32x2_t a) {
- // CHECK-LABEL: test_vcvta_u32_f32
- // CHECK-LABEL: call <2 x i32> @llvm.arm.neon.vcvtau.v2i32.v2f32(<2 x float> %a)
return vcvta_u32_f32(a);
}
+// CHECK-LABEL: define <4 x i32> @test_vcvtaq_s32_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[VCVTAQ_S32_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VCVTAQ_S32_V1_I:%.*]] = call <4 x i32> @llvm.arm.neon.vcvtas.v4i32.v4f32(<4 x float> [[VCVTAQ_S32_V_I]]) #2
+// CHECK: ret <4 x i32> [[VCVTAQ_S32_V1_I]]
int32x4_t test_vcvtaq_s32_f32(float32x4_t a) {
- // CHECK-LABEL: test_vcvtaq_s32_f32
- // CHECK-LABEL: call <4 x i32> @llvm.arm.neon.vcvtas.v4i32.v4f32(<4 x float> %a)
return vcvtaq_s32_f32(a);
}
+// CHECK-LABEL: define <4 x i32> @test_vcvtaq_u32_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[VCVTAQ_U32_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VCVTAQ_U32_V1_I:%.*]] = call <4 x i32> @llvm.arm.neon.vcvtau.v4i32.v4f32(<4 x float> [[VCVTAQ_U32_V_I]]) #2
+// CHECK: ret <4 x i32> [[VCVTAQ_U32_V1_I]]
uint32x4_t test_vcvtaq_u32_f32(float32x4_t a) {
- // CHECK-LABEL: test_vcvtaq_u32_f32
- // CHECK-LABEL: call <4 x i32> @llvm.arm.neon.vcvtau.v4i32.v4f32(<4 x float> %a)
return vcvtaq_u32_f32(a);
}
+// CHECK-LABEL: define <2 x i32> @test_vcvtn_s32_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[VCVTN_S32_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VCVTN_S32_V1_I:%.*]] = call <2 x i32> @llvm.arm.neon.vcvtns.v2i32.v2f32(<2 x float> [[VCVTN_S32_V_I]]) #2
+// CHECK: ret <2 x i32> [[VCVTN_S32_V1_I]]
int32x2_t test_vcvtn_s32_f32(float32x2_t a) {
- // CHECK-LABEL: test_vcvtn_s32_f32
- // CHECK-LABEL: call <2 x i32> @llvm.arm.neon.vcvtns.v2i32.v2f32(<2 x float> %a)
return vcvtn_s32_f32(a);
}
+// CHECK-LABEL: define <2 x i32> @test_vcvtn_u32_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[VCVTN_U32_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VCVTN_U32_V1_I:%.*]] = call <2 x i32> @llvm.arm.neon.vcvtnu.v2i32.v2f32(<2 x float> [[VCVTN_U32_V_I]]) #2
+// CHECK: ret <2 x i32> [[VCVTN_U32_V1_I]]
uint32x2_t test_vcvtn_u32_f32(float32x2_t a) {
- // CHECK-LABEL: test_vcvtn_u32_f32
- // CHECK-LABEL: call <2 x i32> @llvm.arm.neon.vcvtnu.v2i32.v2f32(<2 x float> %a)
return vcvtn_u32_f32(a);
}
+// CHECK-LABEL: define <4 x i32> @test_vcvtnq_s32_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[VCVTNQ_S32_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VCVTNQ_S32_V1_I:%.*]] = call <4 x i32> @llvm.arm.neon.vcvtns.v4i32.v4f32(<4 x float> [[VCVTNQ_S32_V_I]]) #2
+// CHECK: ret <4 x i32> [[VCVTNQ_S32_V1_I]]
int32x4_t test_vcvtnq_s32_f32(float32x4_t a) {
- // CHECK-LABEL: test_vcvtnq_s32_f32
- // CHECK-LABEL: call <4 x i32> @llvm.arm.neon.vcvtns.v4i32.v4f32(<4 x float> %a)
return vcvtnq_s32_f32(a);
}
+// CHECK-LABEL: define <4 x i32> @test_vcvtnq_u32_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[VCVTNQ_U32_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VCVTNQ_U32_V1_I:%.*]] = call <4 x i32> @llvm.arm.neon.vcvtnu.v4i32.v4f32(<4 x float> [[VCVTNQ_U32_V_I]]) #2
+// CHECK: ret <4 x i32> [[VCVTNQ_U32_V1_I]]
uint32x4_t test_vcvtnq_u32_f32(float32x4_t a) {
- // CHECK-LABEL: test_vcvtnq_u32_f32
- // CHECK-LABEL: call <4 x i32> @llvm.arm.neon.vcvtnu.v4i32.v4f32(<4 x float> %a)
return vcvtnq_u32_f32(a);
}
+// CHECK-LABEL: define <2 x i32> @test_vcvtp_s32_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[VCVTP_S32_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VCVTP_S32_V1_I:%.*]] = call <2 x i32> @llvm.arm.neon.vcvtps.v2i32.v2f32(<2 x float> [[VCVTP_S32_V_I]]) #2
+// CHECK: ret <2 x i32> [[VCVTP_S32_V1_I]]
int32x2_t test_vcvtp_s32_f32(float32x2_t a) {
- // CHECK-LABEL: test_vcvtp_s32_f32
- // CHECK-LABEL: call <2 x i32> @llvm.arm.neon.vcvtps.v2i32.v2f32(<2 x float> %a)
return vcvtp_s32_f32(a);
}
+// CHECK-LABEL: define <2 x i32> @test_vcvtp_u32_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[VCVTP_U32_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VCVTP_U32_V1_I:%.*]] = call <2 x i32> @llvm.arm.neon.vcvtpu.v2i32.v2f32(<2 x float> [[VCVTP_U32_V_I]]) #2
+// CHECK: ret <2 x i32> [[VCVTP_U32_V1_I]]
uint32x2_t test_vcvtp_u32_f32(float32x2_t a) {
- // CHECK-LABEL: test_vcvtp_u32_f32
- // CHECK-LABEL: call <2 x i32> @llvm.arm.neon.vcvtpu.v2i32.v2f32(<2 x float> %a)
return vcvtp_u32_f32(a);
}
+// CHECK-LABEL: define <4 x i32> @test_vcvtpq_s32_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[VCVTPQ_S32_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VCVTPQ_S32_V1_I:%.*]] = call <4 x i32> @llvm.arm.neon.vcvtps.v4i32.v4f32(<4 x float> [[VCVTPQ_S32_V_I]]) #2
+// CHECK: ret <4 x i32> [[VCVTPQ_S32_V1_I]]
int32x4_t test_vcvtpq_s32_f32(float32x4_t a) {
- // CHECK-LABEL: test_vcvtpq_s32_f32
- // CHECK-LABEL: call <4 x i32> @llvm.arm.neon.vcvtps.v4i32.v4f32(<4 x float> %a)
return vcvtpq_s32_f32(a);
}
+// CHECK-LABEL: define <4 x i32> @test_vcvtpq_u32_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[VCVTPQ_U32_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VCVTPQ_U32_V1_I:%.*]] = call <4 x i32> @llvm.arm.neon.vcvtpu.v4i32.v4f32(<4 x float> [[VCVTPQ_U32_V_I]]) #2
+// CHECK: ret <4 x i32> [[VCVTPQ_U32_V1_I]]
uint32x4_t test_vcvtpq_u32_f32(float32x4_t a) {
- // CHECK-LABEL: test_vcvtpq_u32_f32
- // CHECK-LABEL: call <4 x i32> @llvm.arm.neon.vcvtpu.v4i32.v4f32(<4 x float> %a)
return vcvtpq_u32_f32(a);
}
+// CHECK-LABEL: define <2 x i32> @test_vcvtm_s32_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[VCVTM_S32_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VCVTM_S32_V1_I:%.*]] = call <2 x i32> @llvm.arm.neon.vcvtms.v2i32.v2f32(<2 x float> [[VCVTM_S32_V_I]]) #2
+// CHECK: ret <2 x i32> [[VCVTM_S32_V1_I]]
int32x2_t test_vcvtm_s32_f32(float32x2_t a) {
- // CHECK-LABEL: test_vcvtm_s32_f32
- // CHECK-LABEL: call <2 x i32> @llvm.arm.neon.vcvtms.v2i32.v2f32(<2 x float> %a)
return vcvtm_s32_f32(a);
}
+// CHECK-LABEL: define <2 x i32> @test_vcvtm_u32_f32(<2 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
+// CHECK: [[VCVTM_U32_V_I:%.*]] = bitcast <8 x i8> [[TMP0]] to <2 x float>
+// CHECK: [[VCVTM_U32_V1_I:%.*]] = call <2 x i32> @llvm.arm.neon.vcvtmu.v2i32.v2f32(<2 x float> [[VCVTM_U32_V_I]]) #2
+// CHECK: ret <2 x i32> [[VCVTM_U32_V1_I]]
uint32x2_t test_vcvtm_u32_f32(float32x2_t a) {
- // CHECK-LABEL: test_vcvtm_u32_f32
- // CHECK-LABEL: call <2 x i32> @llvm.arm.neon.vcvtmu.v2i32.v2f32(<2 x float> %a)
return vcvtm_u32_f32(a);
}
+// CHECK-LABEL: define <4 x i32> @test_vcvtmq_s32_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[VCVTMQ_S32_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VCVTMQ_S32_V1_I:%.*]] = call <4 x i32> @llvm.arm.neon.vcvtms.v4i32.v4f32(<4 x float> [[VCVTMQ_S32_V_I]]) #2
+// CHECK: ret <4 x i32> [[VCVTMQ_S32_V1_I]]
int32x4_t test_vcvtmq_s32_f32(float32x4_t a) {
- // CHECK-LABEL: test_vcvtmq_s32_f32
- // CHECK-LABEL: call <4 x i32> @llvm.arm.neon.vcvtms.v4i32.v4f32(<4 x float> %a)
return vcvtmq_s32_f32(a);
}
+// CHECK-LABEL: define <4 x i32> @test_vcvtmq_u32_f32(<4 x float> %a) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a to <16 x i8>
+// CHECK: [[VCVTMQ_U32_V_I:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VCVTMQ_U32_V1_I:%.*]] = call <4 x i32> @llvm.arm.neon.vcvtmu.v4i32.v4f32(<4 x float> [[VCVTMQ_U32_V_I]]) #2
+// CHECK: ret <4 x i32> [[VCVTMQ_U32_V1_I]]
uint32x4_t test_vcvtmq_u32_f32(float32x4_t a) {
- // CHECK-LABEL: test_vcvtmq_u32_f32
- // CHECK-LABEL: call <4 x i32> @llvm.arm.neon.vcvtmu.v4i32.v4f32(<4 x float> %a)
return vcvtmq_u32_f32(a);
}
Modified: cfe/trunk/test/CodeGen/arm-neon-vget.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm-neon-vget.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/arm-neon-vget.c (original)
+++ cfe/trunk/test/CodeGen/arm-neon-vget.c Wed Mar 9 12:54:42 2016
@@ -1,124 +1,123 @@
-// REQUIRES: arm-registered-target
// RUN: %clang_cc1 -triple thumbv7-apple-darwin \
// RUN: -target-abi apcs-gnu \
// RUN: -target-cpu cortex-a8 \
// RUN: -mfloat-abi soft \
// RUN: -target-feature +soft-float-abi \
// RUN: -ffreestanding \
-// RUN: -emit-llvm -w -O1 -o - %s | FileCheck %s
+// RUN: -emit-llvm -w -o - %s | opt -S -mem2reg | FileCheck %s
#include <arm_neon.h>
// Check that the vget_low/vget_high intrinsics generate a single shuffle
// without any bitcasting.
int8x8_t low_s8(int8x16_t a) {
-// CHECK: shufflevector <16 x i8> %a, <16 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+// CHECK: shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
return vget_low_s8(a);
}
uint8x8_t low_u8 (uint8x16_t a) {
-// CHECK: shufflevector <16 x i8> %a, <16 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+// CHECK: shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
return vget_low_u8(a);
}
int16x4_t low_s16( int16x8_t a) {
-// CHECK: shufflevector <8 x i16> %a, <8 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+// CHECK: shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
return vget_low_s16(a);
}
uint16x4_t low_u16(uint16x8_t a) {
-// CHECK: shufflevector <8 x i16> %a, <8 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+// CHECK: shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
return vget_low_u16(a);
}
int32x2_t low_s32( int32x4_t a) {
-// CHECK: shufflevector <4 x i32> %a, <4 x i32> undef, <2 x i32> <i32 0, i32 1>
+// CHECK: shufflevector <4 x i32> %a, <4 x i32> %a, <2 x i32> <i32 0, i32 1>
return vget_low_s32(a);
}
uint32x2_t low_u32(uint32x4_t a) {
-// CHECK: shufflevector <4 x i32> %a, <4 x i32> undef, <2 x i32> <i32 0, i32 1>
+// CHECK: shufflevector <4 x i32> %a, <4 x i32> %a, <2 x i32> <i32 0, i32 1>
return vget_low_u32(a);
}
int64x1_t low_s64( int64x2_t a) {
-// CHECK: shufflevector <2 x i64> %a, <2 x i64> undef, <1 x i32> zeroinitializer
+// CHECK: shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> zeroinitializer
return vget_low_s64(a);
}
uint64x1_t low_u64(uint64x2_t a) {
-// CHECK: shufflevector <2 x i64> %a, <2 x i64> undef, <1 x i32> zeroinitializer
+// CHECK: shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> zeroinitializer
return vget_low_u64(a);
}
poly8x8_t low_p8 (poly8x16_t a) {
-// CHECK: shufflevector <16 x i8> %a, <16 x i8> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+// CHECK: shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
return vget_low_p8(a);
}
poly16x4_t low_p16(poly16x8_t a) {
-// CHECK: shufflevector <8 x i16> %a, <8 x i16> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+// CHECK: shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
return vget_low_p16(a);
}
float32x2_t low_f32(float32x4_t a) {
-// CHECK: shufflevector <4 x float> %a, <4 x float> undef, <2 x i32> <i32 0, i32 1>
+// CHECK: shufflevector <4 x float> %a, <4 x float> %a, <2 x i32> <i32 0, i32 1>
return vget_low_f32(a);
}
int8x8_t high_s8(int8x16_t a) {
-// CHECK: shufflevector <16 x i8> %a, <16 x i8> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
return vget_high_s8(a);
}
uint8x8_t high_u8 (uint8x16_t a) {
-// CHECK: shufflevector <16 x i8> %a, <16 x i8> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
return vget_high_u8(a);
}
int16x4_t high_s16( int16x8_t a) {
-// CHECK: shufflevector <8 x i16> %a, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+// CHECK: shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
return vget_high_s16(a);
}
uint16x4_t high_u16(uint16x8_t a) {
-// CHECK: shufflevector <8 x i16> %a, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+// CHECK: shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
return vget_high_u16(a);
}
int32x2_t high_s32( int32x4_t a) {
-// CHECK: shufflevector <4 x i32> %a, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
+// CHECK: shufflevector <4 x i32> %a, <4 x i32> %a, <2 x i32> <i32 2, i32 3>
return vget_high_s32(a);
}
uint32x2_t high_u32(uint32x4_t a) {
-// CHECK: shufflevector <4 x i32> %a, <4 x i32> undef, <2 x i32> <i32 2, i32 3>
+// CHECK: shufflevector <4 x i32> %a, <4 x i32> %a, <2 x i32> <i32 2, i32 3>
return vget_high_u32(a);
}
int64x1_t high_s64( int64x2_t a) {
-// CHECK: shufflevector <2 x i64> %a, <2 x i64> undef, <1 x i32> <i32 1>
+// CHECK: shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> <i32 1>
return vget_high_s64(a);
}
uint64x1_t high_u64(uint64x2_t a) {
-// CHECK: shufflevector <2 x i64> %a, <2 x i64> undef, <1 x i32> <i32 1>
+// CHECK: shufflevector <2 x i64> %a, <2 x i64> %a, <1 x i32> <i32 1>
return vget_high_u64(a);
}
poly8x8_t high_p8 (poly8x16_t a) {
-// CHECK: shufflevector <16 x i8> %a, <16 x i8> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+// CHECK: shufflevector <16 x i8> %a, <16 x i8> %a, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
return vget_high_p8(a);
}
poly16x4_t high_p16(poly16x8_t a) {
-// CHECK: shufflevector <8 x i16> %a, <8 x i16> undef, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+// CHECK: shufflevector <8 x i16> %a, <8 x i16> %a, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
return vget_high_p16(a);
}
float32x2_t high_f32(float32x4_t a) {
-// CHECK: shufflevector <4 x float> %a, <4 x float> undef, <2 x i32> <i32 2, i32 3>
+// CHECK: shufflevector <4 x float> %a, <4 x float> %a, <2 x i32> <i32 2, i32 3>
return vget_high_f32(a);
}
Modified: cfe/trunk/test/CodeGen/arm64-be-bitfield.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm64-be-bitfield.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/arm64-be-bitfield.c (original)
+++ cfe/trunk/test/CodeGen/arm64-be-bitfield.c Wed Mar 9 12:54:42 2016
@@ -1,6 +1,4 @@
-// REQUIRES: aarch64-registered-target
// RUN: %clang_cc1 -triple aarch64_be-linux-gnu -ffreestanding -emit-llvm -O0 -o - %s | FileCheck --check-prefix IR %s
-// RUN: %clang_cc1 -triple aarch64_be-linux-gnu -ffreestanding -S -O1 -o - %s | FileCheck --check-prefix ARM %s
struct bt3 { signed b2:10; signed b3:10; } b16;
@@ -10,6 +8,5 @@ signed callee_b0f(struct bt3 bp11) {
// IR: store i64 [[ARG]], i64* [[PTR:%.*]], align 8
// IR: [[BITCAST:%.*]] = bitcast i64* [[PTR]] to i8*
// IR: call void @llvm.memcpy.p0i8.p0i8.i64(i8* {{.*}}, i8* [[BITCAST]], i64 4
-// ARM: asr x0, x0, #54
return bp11.b2;
}
Modified: cfe/trunk/test/CodeGen/arm64-crc32.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm64-crc32.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/arm64-crc32.c (original)
+++ cfe/trunk/test/CodeGen/arm64-crc32.c Wed Mar 9 12:54:42 2016
@@ -1,6 +1,6 @@
// REQUIRES: aarch64-registered-target
// RUN: %clang_cc1 -triple arm64-none-linux-gnu \
-// RUN: -O3 -S -emit-llvm -o - %s | FileCheck %s
+// RUN: -S -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s
int crc32b(int a, char b)
{
Modified: cfe/trunk/test/CodeGen/arm64-lanes.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm64-lanes.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/arm64-lanes.c (original)
+++ cfe/trunk/test/CodeGen/arm64-lanes.c Wed Mar 9 12:54:42 2016
@@ -1,74 +1,127 @@
-// RUN: %clang_cc1 -O3 -triple arm64-apple-ios7 -target-feature +neon -ffreestanding -emit-llvm -o - %s | FileCheck %s
-// RUN: %clang_cc1 -O3 -triple aarch64_be-linux-gnu -target-feature +neon -ffreestanding -emit-llvm -o - %s | FileCheck %s --check-prefix CHECK-BE
+// RUN: %clang_cc1 -triple arm64-apple-ios7 -target-feature +neon -ffreestanding -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64_be-linux-gnu -target-feature +neon -ffreestanding -emit-llvm -o - %s | opt -S -mem2reg | FileCheck %s --check-prefix CHECK-BE
#include <arm_neon.h>
-// CHECK-LABEL: @test_vdupb_lane_s8
int8_t test_vdupb_lane_s8(int8x8_t src) {
return vdupb_lane_s8(src, 2);
+ // CHECK-LABEL: @test_vdupb_lane_s8
// CHECK: extractelement <8 x i8> %src, i32 2
- // CHECK-BE: extractelement <8 x i8> %src, i32 5
+
+ // CHECK-BE-LABEL: @test_vdupb_lane_s8
+ // CHECK-BE: [[REV:%.*]] = shufflevector <8 x i8> {{.*}}, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+ // CHECK-BE: extractelement <8 x i8> [[REV]], i32 2
}
-// CHECK-LABEL: @test_vdupb_lane_u8
uint8_t test_vdupb_lane_u8(uint8x8_t src) {
return vdupb_lane_u8(src, 2);
+ // CHECK-LABEL: @test_vdupb_lane_u8
// CHECK: extractelement <8 x i8> %src, i32 2
- // CHECK-BE: extractelement <8 x i8> %src, i32 5
+
+ // CHECK-BE-LABEL: @test_vdupb_lane_u8
+ // CHECK-BE: [[REV:%.*]] = shufflevector <8 x i8> {{.*}}, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+ // CHECK-BE: extractelement <8 x i8> [[REV]], i32 2
}
-// CHECK-LABEL: @test_vduph_lane_s16
int16_t test_vduph_lane_s16(int16x4_t src) {
return vduph_lane_s16(src, 2);
- // CHECK: extractelement <4 x i16> %src, i32 2
- // CHECK-BE: extractelement <4 x i16> %src, i32 1
+ // CHECK-LABEL: @test_vduph_lane_s16
+ // CHECK: [[TMP1:%.*]] = bitcast <4 x i16> %src to [[TYPE:.*]]
+ // CHECK: [[TMP2:%.*]] = bitcast [[TYPE]] [[TMP1]] to <4 x i16>
+ // CHECK: extractelement <4 x i16> [[TMP2]], i32 2
+
+ // CHECK-BE-LABEL: @test_vduph_lane_s16
+ // CHECK-BE: [[REV:%.*]] = shufflevector <4 x i16> {{.*}}, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+ // CHECK-BE: [[TMP1:%.*]] = bitcast <4 x i16> [[REV]] to [[TYPE:.*]]
+ // CHECK-BE: [[TMP2:%.*]] = bitcast [[TYPE]] [[TMP1]] to <4 x i16>
+ // CHECK-BE: extractelement <4 x i16> [[TMP2]], i32 2
}
-// CHECK-LABEL: @test_vduph_lane_u16
uint16_t test_vduph_lane_u16(uint16x4_t src) {
return vduph_lane_u16(src, 2);
- // CHECK: extractelement <4 x i16> %src, i32 2
- // CHECK-BE: extractelement <4 x i16> %src, i32 1
+ // CHECK-LABEL: @test_vduph_lane_u16
+ // CHECK: [[TMP1:%.*]] = bitcast <4 x i16> %src to [[TYPE:.*]]
+ // CHECK: [[TMP2:%.*]] = bitcast [[TYPE]] [[TMP1]] to <4 x i16>
+ // CHECK: extractelement <4 x i16> [[TMP2]], i32 2
+
+ // CHECK-BE-LABEL: @test_vduph_lane_u16
+ // CHECK-BE: [[REV:%.*]] = shufflevector <4 x i16> {{.*}}, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
+ // CHECK-BE: [[TMP1:%.*]] = bitcast <4 x i16> [[REV]] to [[TYPE:.*]]
+ // CHECK-BE: [[TMP2:%.*]] = bitcast [[TYPE]] [[TMP1]] to <4 x i16>
+ // CHECK-BE: extractelement <4 x i16> [[TMP2]], i32 2
}
-// CHECK-LABEL: @test_vdups_lane_s32
int32_t test_vdups_lane_s32(int32x2_t src) {
return vdups_lane_s32(src, 0);
- // CHECK: extractelement <2 x i32> %src, i32 0
- // CHECK-BE: extractelement <2 x i32> %src, i32 1
+ // CHECK-LABEL: @test_vdups_lane_s32
+ // CHECK: [[TMP1:%.*]] = bitcast <2 x i32> %src to [[TYPE:.*]]
+ // CHECK: [[TMP2:%.*]] = bitcast [[TYPE]] [[TMP1]] to <2 x i32>
+ // CHECK: extractelement <2 x i32> [[TMP2]], i32 0
+
+ // CHECK-BE-LABEL: @test_vdups_lane_s32
+ // CHECK-BE: [[REV:%.*]] = shufflevector <2 x i32> {{.*}}, <2 x i32> <i32 1, i32 0>
+ // CHECK-BE: [[TMP1:%.*]] = bitcast <2 x i32> [[REV]] to [[TYPE:.*]]
+ // CHECK-BE: [[TMP2:%.*]] = bitcast [[TYPE]] [[TMP1]] to <2 x i32>
+ // CHECK-BE: extractelement <2 x i32> [[TMP2]], i32 0
}
-// CHECK-LABEL: @test_vdups_lane_u32
uint32_t test_vdups_lane_u32(uint32x2_t src) {
return vdups_lane_u32(src, 0);
- // CHECK: extractelement <2 x i32> %src, i32 0
- // CHECK-BE: extractelement <2 x i32> %src, i32 1
+ // CHECK-LABEL: @test_vdups_lane_u32
+ // CHECK: [[TMP1:%.*]] = bitcast <2 x i32> %src to [[TYPE:.*]]
+ // CHECK: [[TMP2:%.*]] = bitcast [[TYPE]] [[TMP1]] to <2 x i32>
+ // CHECK: extractelement <2 x i32> [[TMP2]], i32 0
+
+ // CHECK-BE-LABEL: @test_vdups_lane_u32
+ // CHECK-BE: [[REV:%.*]] = shufflevector <2 x i32> {{.*}}, <2 x i32> <i32 1, i32 0>
+ // CHECK-BE: [[TMP1:%.*]] = bitcast <2 x i32> [[REV]] to [[TYPE:.*]]
+ // CHECK-BE: [[TMP2:%.*]] = bitcast [[TYPE]] [[TMP1]] to <2 x i32>
+ // CHECK-BE: extractelement <2 x i32> [[TMP2]], i32 0
}
-// CHECK-LABEL: @test_vdups_lane_f32
float32_t test_vdups_lane_f32(float32x2_t src) {
return vdups_lane_f32(src, 0);
- // CHECK: extractelement <2 x float> %src, i32 0
- // CHECK-BE: extractelement <2 x float> %src, i32 1
+ // CHECK-LABEL: @test_vdups_lane_f32
+ // CHECK: [[TMP1:%.*]] = bitcast <2 x float> %src to [[TYPE:.*]]
+ // CHECK: [[TMP2:%.*]] = bitcast [[TYPE]] [[TMP1]] to <2 x float>
+ // CHECK: extractelement <2 x float> [[TMP2]], i32 0
+
+ // CHECK-BE-LABEL: @test_vdups_lane_f32
+ // CHECK-BE: [[REV:%.*]] = shufflevector <2 x float> {{.*}}, <2 x i32> <i32 1, i32 0>
+ // CHECK-BE: [[TMP1:%.*]] = bitcast <2 x float> [[REV]] to [[TYPE:.*]]
+ // CHECK-BE: [[TMP2:%.*]] = bitcast [[TYPE]] [[TMP1]] to <2 x float>
+ // CHECK-BE: extractelement <2 x float> [[TMP2]], i32 0
}
-// CHECK-LABEL: @test_vdupd_lane_s64
int64_t test_vdupd_lane_s64(int64x1_t src) {
return vdupd_lane_s64(src, 0);
- // CHECK: extractelement <1 x i64> %src, i32 0
- // CHECK-BE: extractelement <1 x i64> %src, i32 0
+ // CHECK-LABEL: @test_vdupd_lane_s64
+ // CHECK: [[TMP1:%.*]] = bitcast <1 x i64> %src to [[TYPE:.*]]
+ // CHECK: [[TMP2:%.*]] = bitcast [[TYPE]] [[TMP1]] to <1 x i64>
+ // CHECK: extractelement <1 x i64> [[TMP2]], i32 0
+
+ // CHECK-BE-LABEL: @test_vdupd_lane_s64
+ // CHECK-BE: extractelement <1 x i64> {{.*}}, i32 0
}
-// CHECK-LABEL: @test_vdupd_lane_u64
uint64_t test_vdupd_lane_u64(uint64x1_t src) {
return vdupd_lane_u64(src, 0);
- // CHECK: extractelement <1 x i64> %src, i32 0
- // CHECK-BE: extractelement <1 x i64> %src, i32 0
+ // CHECK-LABEL: @test_vdupd_lane_u64
+ // CHECK: [[TMP1:%.*]] = bitcast <1 x i64> %src to [[TYPE:.*]]
+ // CHECK: [[TMP2:%.*]] = bitcast [[TYPE]] [[TMP1]] to <1 x i64>
+ // CHECK: extractelement <1 x i64> [[TMP2]], i32 0
+
+ // CHECK-BE-LABEL: @test_vdupd_lane_u64
+ // CHECK-BE: extractelement <1 x i64> {{.*}}, i32 0
}
-// CHECK-LABEL: @test_vdupd_lane_f64
float64_t test_vdupd_lane_f64(float64x1_t src) {
return vdupd_lane_f64(src, 0);
- // CHECK: extractelement <1 x double> %src, i32 0
- // CHECK-BE: extractelement <1 x double> %src, i32 0
+ // CHECK-LABEL: @test_vdupd_lane_f64
+ // CHECK: [[TMP1:%.*]] = bitcast <1 x double> %src to [[TYPE:.*]]
+ // CHECK: [[TMP2:%.*]] = bitcast [[TYPE]] [[TMP1]] to <1 x double>
+ // CHECK: extractelement <1 x double> [[TMP2]], i32 0
+
+ // CHECK-BE-LABEL: @test_vdupd_lane_f64
+ // CHECK-BE: extractelement <1 x double> {{.*}}, i32 0
}
Modified: cfe/trunk/test/CodeGen/arm64_vcopy.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm64_vcopy.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/arm64_vcopy.c (original)
+++ cfe/trunk/test/CodeGen/arm64_vcopy.c Wed Mar 9 12:54:42 2016
@@ -1,69 +1,121 @@
-// RUN: %clang_cc1 -O1 -triple arm64-apple-ios7 -target-feature +neon -ffreestanding -S -o - -emit-llvm %s | FileCheck %s
+// RUN: %clang_cc1 -triple arm64-apple-ios7 -target-feature +neon -ffreestanding -S -o - -emit-llvm %s | opt -S -mem2reg | FileCheck %s
// Test ARM64 SIMD copy vector element to vector element: vcopyq_lane*
#include <arm_neon.h>
+// CHECK-LABEL: define <16 x i8> @test_vcopyq_laneq_s8(<16 x i8> %a1, <16 x i8> %a2) #0 {
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <16 x i8> %a2, i32 13
+// CHECK: [[VSET_LANE:%.*]] = insertelement <16 x i8> %a1, i8 [[VGETQ_LANE]], i32 3
+// CHECK: ret <16 x i8> [[VSET_LANE]]
int8x16_t test_vcopyq_laneq_s8(int8x16_t a1, int8x16_t a2) {
- // CHECK-LABEL: test_vcopyq_laneq_s8
return vcopyq_laneq_s8(a1, (int64_t) 3, a2, (int64_t) 13);
- // CHECK: shufflevector <16 x i8> %a1, <16 x i8> %a2, <16 x i32> <i32 0, i32 1, i32 2, i32 29, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
}
+// CHECK-LABEL: define <16 x i8> @test_vcopyq_laneq_u8(<16 x i8> %a1, <16 x i8> %a2) #0 {
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <16 x i8> %a2, i32 13
+// CHECK: [[VSET_LANE:%.*]] = insertelement <16 x i8> %a1, i8 [[VGETQ_LANE]], i32 3
+// CHECK: ret <16 x i8> [[VSET_LANE]]
uint8x16_t test_vcopyq_laneq_u8(uint8x16_t a1, uint8x16_t a2) {
- // CHECK-LABEL: test_vcopyq_laneq_u8
return vcopyq_laneq_u8(a1, (int64_t) 3, a2, (int64_t) 13);
- // CHECK: shufflevector <16 x i8> %a1, <16 x i8> %a2, <16 x i32> <i32 0, i32 1, i32 2, i32 29, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
}
+// CHECK-LABEL: define <8 x i16> @test_vcopyq_laneq_s16(<8 x i16> %a1, <8 x i16> %a2) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a2 to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <8 x i16> [[TMP1]], i32 7
+// CHECK: [[TMP2:%.*]] = bitcast <8 x i16> %a1 to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP2]] to <8 x i16>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <8 x i16> [[TMP3]], i16 [[VGETQ_LANE]], i32 3
+// CHECK: ret <8 x i16> [[VSET_LANE]]
int16x8_t test_vcopyq_laneq_s16(int16x8_t a1, int16x8_t a2) {
- // CHECK-LABEL: test_vcopyq_laneq_s16
return vcopyq_laneq_s16(a1, (int64_t) 3, a2, (int64_t) 7);
- // CHECK: shufflevector <8 x i16> %a1, <8 x i16> %a2, <8 x i32> <i32 0, i32 1, i32 2, i32 15, i32 4, i32 5, i32 6, i32 7>
}
+// CHECK-LABEL: define <8 x i16> @test_vcopyq_laneq_u16(<8 x i16> %a1, <8 x i16> %a2) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <8 x i16> %a2 to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <8 x i16>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <8 x i16> [[TMP1]], i32 7
+// CHECK: [[TMP2:%.*]] = bitcast <8 x i16> %a1 to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP2]] to <8 x i16>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <8 x i16> [[TMP3]], i16 [[VGETQ_LANE]], i32 3
+// CHECK: ret <8 x i16> [[VSET_LANE]]
uint16x8_t test_vcopyq_laneq_u16(uint16x8_t a1, uint16x8_t a2) {
- // CHECK-LABEL: test_vcopyq_laneq_u16
return vcopyq_laneq_u16(a1, (int64_t) 3, a2, (int64_t) 7);
- // CHECK: shufflevector <8 x i16> %a1, <8 x i16> %a2, <8 x i32> <i32 0, i32 1, i32 2, i32 15, i32 4, i32 5, i32 6, i32 7>
}
+// CHECK-LABEL: define <4 x i32> @test_vcopyq_laneq_s32(<4 x i32> %a1, <4 x i32> %a2) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a2 to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <4 x i32> [[TMP1]], i32 3
+// CHECK: [[TMP2:%.*]] = bitcast <4 x i32> %a1 to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP2]] to <4 x i32>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <4 x i32> [[TMP3]], i32 [[VGETQ_LANE]], i32 3
+// CHECK: ret <4 x i32> [[VSET_LANE]]
int32x4_t test_vcopyq_laneq_s32(int32x4_t a1, int32x4_t a2) {
- // CHECK-LABEL: test_vcopyq_laneq_s32
return vcopyq_laneq_s32(a1, (int64_t) 3, a2, (int64_t) 3);
- // CHECK: shufflevector <4 x i32> %a1, <4 x i32> %a2, <4 x i32> <i32 0, i32 1, i32 2, i32 7>
}
+// CHECK-LABEL: define <4 x i32> @test_vcopyq_laneq_u32(<4 x i32> %a1, <4 x i32> %a2) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x i32> %a2 to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x i32>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <4 x i32> [[TMP1]], i32 3
+// CHECK: [[TMP2:%.*]] = bitcast <4 x i32> %a1 to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP2]] to <4 x i32>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <4 x i32> [[TMP3]], i32 [[VGETQ_LANE]], i32 3
+// CHECK: ret <4 x i32> [[VSET_LANE]]
uint32x4_t test_vcopyq_laneq_u32(uint32x4_t a1, uint32x4_t a2) {
- // CHECK-LABEL: test_vcopyq_laneq_u32
return vcopyq_laneq_u32(a1, (int64_t) 3, a2, (int64_t) 3);
- // CHECK: shufflevector <4 x i32> %a1, <4 x i32> %a2, <4 x i32> <i32 0, i32 1, i32 2, i32 7>
}
+// CHECK-LABEL: define <2 x i64> @test_vcopyq_laneq_s64(<2 x i64> %a1, <2 x i64> %a2) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a2 to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <2 x i64> [[TMP1]], i32 1
+// CHECK: [[TMP2:%.*]] = bitcast <2 x i64> %a1 to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP2]] to <2 x i64>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <2 x i64> [[TMP3]], i64 [[VGETQ_LANE]], i32 0
+// CHECK: ret <2 x i64> [[VSET_LANE]]
int64x2_t test_vcopyq_laneq_s64(int64x2_t a1, int64x2_t a2) {
- // CHECK-LABEL: test_vcopyq_laneq_s64
return vcopyq_laneq_s64(a1, (int64_t) 0, a2, (int64_t) 1);
- // CHECK: shufflevector <2 x i64> %a1, <2 x i64> %a2, <2 x i32> <i32 3, i32 1>
}
+// CHECK-LABEL: define <2 x i64> @test_vcopyq_laneq_u64(<2 x i64> %a1, <2 x i64> %a2) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x i64> %a2 to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x i64>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <2 x i64> [[TMP1]], i32 1
+// CHECK: [[TMP2:%.*]] = bitcast <2 x i64> %a1 to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP2]] to <2 x i64>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <2 x i64> [[TMP3]], i64 [[VGETQ_LANE]], i32 0
+// CHECK: ret <2 x i64> [[VSET_LANE]]
uint64x2_t test_vcopyq_laneq_u64(uint64x2_t a1, uint64x2_t a2) {
- // CHECK-LABEL: test_vcopyq_laneq_u64
return vcopyq_laneq_u64(a1, (int64_t) 0, a2, (int64_t) 1);
- // CHECK: shufflevector <2 x i64> %a1, <2 x i64> %a2, <2 x i32> <i32 3, i32 1>
}
+// CHECK-LABEL: define <4 x float> @test_vcopyq_laneq_f32(<4 x float> %a1, <4 x float> %a2) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <4 x float> %a2 to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <4 x float>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <4 x float> [[TMP1]], i32 3
+// CHECK: [[TMP2:%.*]] = bitcast <4 x float> %a1 to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP2]] to <4 x float>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <4 x float> [[TMP3]], float [[VGETQ_LANE]], i32 0
+// CHECK: ret <4 x float> [[VSET_LANE]]
float32x4_t test_vcopyq_laneq_f32(float32x4_t a1, float32x4_t a2) {
- // CHECK-LABEL: test_vcopyq_laneq_f32
return vcopyq_laneq_f32(a1, 0, a2, 3);
- // CHECK: shufflevector <4 x float> %a1, <4 x float> %a2, <4 x i32> <i32 7, i32 1, i32 2, i32 3>
}
+// CHECK-LABEL: define <2 x double> @test_vcopyq_laneq_f64(<2 x double> %a1, <2 x double> %a2) #0 {
+// CHECK: [[TMP0:%.*]] = bitcast <2 x double> %a2 to <16 x i8>
+// CHECK: [[TMP1:%.*]] = bitcast <16 x i8> [[TMP0]] to <2 x double>
+// CHECK: [[VGETQ_LANE:%.*]] = extractelement <2 x double> [[TMP1]], i32 1
+// CHECK: [[TMP2:%.*]] = bitcast <2 x double> %a1 to <16 x i8>
+// CHECK: [[TMP3:%.*]] = bitcast <16 x i8> [[TMP2]] to <2 x double>
+// CHECK: [[VSET_LANE:%.*]] = insertelement <2 x double> [[TMP3]], double [[VGETQ_LANE]], i32 0
+// CHECK: ret <2 x double> [[VSET_LANE]]
float64x2_t test_vcopyq_laneq_f64(float64x2_t a1, float64x2_t a2) {
- // CHECK-LABEL: test_vcopyq_laneq_f64
return vcopyq_laneq_f64(a1, 0, a2, 1);
- // CHECK: shufflevector <2 x double> %a1, <2 x double> %a2, <2 x i32> <i32 3, i32 1>
}
Modified: cfe/trunk/test/CodeGen/arm64_vcreate.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm64_vcreate.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/arm64_vcreate.c (original)
+++ cfe/trunk/test/CodeGen/arm64_vcreate.c Wed Mar 9 12:54:42 2016
@@ -1,7 +1,6 @@
-// RUN: %clang_cc1 -O1 -triple arm64-apple-ios7 -target-feature +neon -ffreestanding -S -o - -emit-llvm %s | FileCheck %s
+// RUN: %clang_cc1 -triple arm64-apple-ios7 -target-feature +neon -ffreestanding -S -o - -emit-llvm %s | opt -S -mem2reg | FileCheck %s
// Test ARM64 SIMD vcreate intrinsics
-/*#include <arm_neon.h>*/
#include <arm_neon.h>
float32x2_t test_vcreate_f32(uint64_t a1) {
@@ -10,14 +9,3 @@ float32x2_t test_vcreate_f32(uint64_t a1
// CHECK: bitcast {{.*}} to <2 x float>
// CHECK-NEXT: ret
}
-
-// FIXME enable when scalar_to_vector in backend is fixed. Also, change
-// CHECK@ to CHECK<colon> and CHECK-NEXT@ to CHECK-NEXT<colon>
-/*
-float64x1_t test_vcreate_f64(uint64_t a1) {
- // CHECK@ test_vcreate_f64
- return vcreate_f64(a1);
- // CHECK@ llvm.aarch64.neon.saddlv.i64.v2i32
- // CHECK-NEXT@ ret
-}
-*/
Modified: cfe/trunk/test/CodeGen/arm64_vdupq_n_f64.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/arm64_vdupq_n_f64.c?rev=263048&r1=263047&r2=263048&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/arm64_vdupq_n_f64.c (original)
+++ cfe/trunk/test/CodeGen/arm64_vdupq_n_f64.c Wed Mar 9 12:54:42 2016
@@ -1,88 +1,78 @@
-// RUN: %clang_cc1 -O3 -triple arm64-apple-ios7 -target-feature +neon -ffreestanding -S -o - %s | FileCheck %s
-// RUN: %clang_cc1 -O3 -triple arm64-apple-ios7 -target-feature +neon -ffreestanding -S -o - -emit-llvm %s | \
-// RUN: FileCheck -check-prefix=CHECK-IR %s
-// REQUIRES: aarch64-registered-target
-
-/// Test vdupq_n_f64 and vmovq_nf64 ARM64 intrinsics
-// <rdar://problem/11778405> ARM64: vdupq_n_f64 and vdupq_lane_f64 intrinsics
-// missing
-
+// RUN: %clang_cc1 -triple arm64-apple-ios7 -target-feature +neon -ffreestanding -fallow-half-arguments-and-returns -S -o - -emit-llvm %s | opt -S -mem2reg | FileCheck %s
#include <arm_neon.h>
// vdupq_n_f64 -> dup.2d v0, v0[0]
//
-float64x2_t test_vdupq_n_f64(float64_t w)
-{
+// CHECK-LABEL: define <2 x double> @test_vdupq_n_f64(double %w) #0 {
+// CHECK: [[VECINIT_I:%.*]] = insertelement <2 x double> undef, double %w, i32 0
+// CHECK: [[VECINIT1_I:%.*]] = insertelement <2 x double> [[VECINIT_I]], double %w, i32 1
+// CHECK: ret <2 x double> [[VECINIT1_I]]
+float64x2_t test_vdupq_n_f64(float64_t w) {
return vdupq_n_f64(w);
- // CHECK-LABEL: test_vdupq_n_f64:
- // CHECK: dup.2d v0, v0[0]
- // CHECK-NEXT: ret
}
// might as well test this while we're here
// vdupq_n_f32 -> dup.4s v0, v0[0]
-float32x4_t test_vdupq_n_f32(float32_t w)
-{
+// CHECK-LABEL: define <4 x float> @test_vdupq_n_f32(float %w) #0 {
+// CHECK: [[VECINIT_I:%.*]] = insertelement <4 x float> undef, float %w, i32 0
+// CHECK: [[VECINIT1_I:%.*]] = insertelement <4 x float> [[VECINIT_I]], float %w, i32 1
+// CHECK: [[VECINIT2_I:%.*]] = insertelement <4 x float> [[VECINIT1_I]], float %w, i32 2
+// CHECK: [[VECINIT3_I:%.*]] = insertelement <4 x float> [[VECINIT2_I]], float %w, i32 3
+// CHECK: ret <4 x float> [[VECINIT3_I]]
+float32x4_t test_vdupq_n_f32(float32_t w) {
return vdupq_n_f32(w);
- // CHECK-LABEL: test_vdupq_n_f32:
- // CHECK: dup.4s v0, v0[0]
- // CHECK-NEXT: ret
}
// vdupq_lane_f64 -> dup.2d v0, v0[0]
// this was in <rdar://problem/11778405>, but had already been implemented,
// test anyway
-float64x2_t test_vdupq_lane_f64(float64x1_t V)
-{
+// CHECK-LABEL: define <2 x double> @test_vdupq_lane_f64(<1 x double> %V) #0 {
+// CHECK: [[SHUFFLE:%.*]] = shufflevector <1 x double> %V, <1 x double> %V, <2 x i32> zeroinitializer
+// CHECK: ret <2 x double> [[SHUFFLE]]
+float64x2_t test_vdupq_lane_f64(float64x1_t V) {
return vdupq_lane_f64(V, 0);
- // CHECK-LABEL: test_vdupq_lane_f64:
- // CHECK: dup.2d v0, v0[0]
- // CHECK-NEXT: ret
}
// vmovq_n_f64 -> dup Vd.2d,X0
// this wasn't in <rdar://problem/11778405>, but it was between the vdups
-float64x2_t test_vmovq_n_f64(float64_t w)
-{
+// CHECK-LABEL: define <2 x double> @test_vmovq_n_f64(double %w) #0 {
+// CHECK: [[VECINIT_I:%.*]] = insertelement <2 x double> undef, double %w, i32 0
+// CHECK: [[VECINIT1_I:%.*]] = insertelement <2 x double> [[VECINIT_I]], double %w, i32 1
+// CHECK: ret <2 x double> [[VECINIT1_I]]
+float64x2_t test_vmovq_n_f64(float64_t w) {
return vmovq_n_f64(w);
- // CHECK-LABEL: test_vmovq_n_f64:
- // CHECK: dup.2d v0, v0[0]
- // CHECK-NEXT: ret
}
-float16x4_t test_vmov_n_f16(float16_t *a1)
-{
- // CHECK-IR-LABEL: test_vmov_n_f16
+// CHECK-LABEL: define <4 x half> @test_vmov_n_f16(half* %a1) #0 {
+// CHECK: [[TMP0:%.*]] = load half, half* %a1, align 2
+// CHECK: [[VECINIT:%.*]] = insertelement <4 x half> undef, half [[TMP0]], i32 0
+// CHECK: [[VECINIT1:%.*]] = insertelement <4 x half> [[VECINIT]], half [[TMP0]], i32 1
+// CHECK: [[VECINIT2:%.*]] = insertelement <4 x half> [[VECINIT]]1, half [[TMP0]], i32 2
+// CHECK: [[VECINIT3:%.*]] = insertelement <4 x half> [[VECINIT]]2, half [[TMP0]], i32 3
+// CHECK: ret <4 x half> [[VECINIT]]3
+float16x4_t test_vmov_n_f16(float16_t *a1) {
return vmov_n_f16(*a1);
- // CHECK-IR: insertelement {{.*}} i32 0{{ *$}}
- // CHECK-IR: insertelement {{.*}} i32 1{{ *$}}
- // CHECK-IR: insertelement {{.*}} i32 2{{ *$}}
- // CHECK-IR: insertelement {{.*}} i32 3{{ *$}}
}
-// Disable until scalar problem in backend is fixed. Change CHECK-IR@ to
-// CHECK-IR<colon>
/*
-float64x1_t test_vmov_n_f64(float64_t a1)
-{
- // CHECK-IR@ test_vmov_n_f64
+float64x1_t test_vmov_n_f64(float64_t a1) {
return vmov_n_f64(a1);
- // CHECK-IR@ insertelement {{.*}} i32 0{{ *$}}
}
*/
-float16x8_t test_vmovq_n_f16(float16_t *a1)
-{
- // CHECK-IR-LABEL: test_vmovq_n_f16
+// CHECK-LABEL: define <8 x half> @test_vmovq_n_f16(half* %a1) #0 {
+// CHECK: [[TMP0:%.*]] = load half, half* %a1, align 2
+// CHECK: [[VECINIT:%.*]] = insertelement <8 x half> undef, half [[TMP0]], i32 0
+// CHECK: [[VECINIT1:%.*]] = insertelement <8 x half> [[VECINIT]], half [[TMP0]], i32 1
+// CHECK: [[VECINIT2:%.*]] = insertelement <8 x half> [[VECINIT]]1, half [[TMP0]], i32 2
+// CHECK: [[VECINIT3:%.*]] = insertelement <8 x half> [[VECINIT]]2, half [[TMP0]], i32 3
+// CHECK: [[VECINIT4:%.*]] = insertelement <8 x half> [[VECINIT]]3, half [[TMP0]], i32 4
+// CHECK: [[VECINIT5:%.*]] = insertelement <8 x half> [[VECINIT]]4, half [[TMP0]], i32 5
+// CHECK: [[VECINIT6:%.*]] = insertelement <8 x half> [[VECINIT]]5, half [[TMP0]], i32 6
+// CHECK: [[VECINIT7:%.*]] = insertelement <8 x half> [[VECINIT]]6, half [[TMP0]], i32 7
+// CHECK: ret <8 x half> [[VECINIT]]7
+float16x8_t test_vmovq_n_f16(float16_t *a1) {
return vmovq_n_f16(*a1);
- // CHECK-IR: insertelement {{.*}} i32 0{{ *$}}
- // CHECK-IR: insertelement {{.*}} i32 1{{ *$}}
- // CHECK-IR: insertelement {{.*}} i32 2{{ *$}}
- // CHECK-IR: insertelement {{.*}} i32 3{{ *$}}
- // CHECK-IR: insertelement {{.*}} i32 4{{ *$}}
- // CHECK-IR: insertelement {{.*}} i32 5{{ *$}}
- // CHECK-IR: insertelement {{.*}} i32 6{{ *$}}
- // CHECK-IR: insertelement {{.*}} i32 7{{ *$}}
}
More information about the cfe-commits
mailing list