r195073 - Implement AArch64 SISD intrinsics for vget_high and vget_low.
Jiangning Liu
jiangning.liu at arm.com
Mon Nov 18 17:46:34 PST 2013
Author: jiangning
Date: Mon Nov 18 19:46:34 2013
New Revision: 195073
URL: http://llvm.org/viewvc/llvm-project?rev=195073&view=rev
Log:
Implement AArch64 SISD intrinsics for vget_high and vget_low.
Added:
cfe/trunk/test/CodeGen/aarch64-neon-vget-hilo.c
Modified:
cfe/trunk/include/clang/Basic/arm_neon.td
Modified: cfe/trunk/include/clang/Basic/arm_neon.td
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Basic/arm_neon.td?rev=195073&r1=195072&r2=195073&view=diff
==============================================================================
--- cfe/trunk/include/clang/Basic/arm_neon.td (original)
+++ cfe/trunk/include/clang/Basic/arm_neon.td Mon Nov 18 19:46:34 2013
@@ -979,6 +979,13 @@ def SCALAR_SUB : SInst<"vsub", "sss", "
// Scalar Saturating Sub
def SCALAR_QSUB : SInst<"vqsub", "sss", "ScSsSiSlSUcSUsSUiSUl">;
+let InstName = "vmov" in {
+def VGET_HIGH_A64 : NoTestOpInst<"vget_high", "dk", "csilhfdUcUsUiUlPcPsPl",
+ OP_HI>;
+def VGET_LOW_A64 : NoTestOpInst<"vget_low", "dk", "csilhfdUcUsUiUlPcPsPl",
+ OP_LO>;
+}
+
////////////////////////////////////////////////////////////////////////////////
// Scalar Shift
// Scalar Shift Left
Added: 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=195073&view=auto
==============================================================================
--- cfe/trunk/test/CodeGen/aarch64-neon-vget-hilo.c (added)
+++ cfe/trunk/test/CodeGen/aarch64-neon-vget-hilo.c Mon Nov 18 19:46:34 2013
@@ -0,0 +1,176 @@
+// REQUIRES: aarch64-registered-target
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +neon \
+// RUN: -ffp-contract=fast -S -O3 -o - %s | FileCheck %s
+
+// Test new aarch64 intrinsics and types
+
+#include <arm_neon.h>
+
+int8x8_t test_vget_high_s8(int8x16_t a) {
+ // CHECK: test_vget_high_s8
+ return vget_high_s8(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+int16x4_t test_vget_high_s16(int16x8_t a) {
+ // CHECK: test_vget_high_s16
+ return vget_high_s16(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+int32x2_t test_vget_high_s32(int32x4_t a) {
+ // CHECK: test_vget_high_s32
+ return vget_high_s32(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+int64x1_t test_vget_high_s64(int64x2_t a) {
+ // CHECK: test_vget_high_s64
+ return vget_high_s64(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+uint8x8_t test_vget_high_u8(uint8x16_t a) {
+ // CHECK: test_vget_high_u8
+ return vget_high_u8(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+uint16x4_t test_vget_high_u16(uint16x8_t a) {
+ // CHECK: test_vget_high_u16
+ return vget_high_u16(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+uint32x2_t test_vget_high_u32(uint32x4_t a) {
+ // CHECK: test_vget_high_u32
+ return vget_high_u32(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+uint64x1_t test_vget_high_u64(uint64x2_t a) {
+ // CHECK: test_vget_high_u64
+ return vget_high_u64(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+poly64x1_t test_vget_high_p64(poly64x2_t a) {
+ // CHECK: test_vget_high_p64
+ return vget_high_p64(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+float16x4_t test_vget_high_f16(float16x8_t a) {
+ // CHECK: test_vget_high_f16
+ return vget_high_f16(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+float32x2_t test_vget_high_f32(float32x4_t a) {
+ // CHECK: test_vget_high_f32
+ return vget_high_f32(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+poly8x8_t test_vget_high_p8(poly8x16_t a) {
+ // CHECK: test_vget_high_p8
+ return vget_high_p8(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+poly16x4_t test_vget_high_p16(poly16x8_t a) {
+ // CHECK: test_vget_high_p16
+ return vget_high_p16(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+float64x1_t test_vget_high_f64(float64x2_t a) {
+ // CHECK: test_vget_high_f64
+ return vget_high_f64(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[1]
+}
+
+int8x8_t test_vget_low_s8(int8x16_t a) {
+ // CHECK: test_vget_low_s8
+ return vget_low_s8(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+int16x4_t test_vget_low_s16(int16x8_t a) {
+ // CHECK: test_vget_low_s16
+ return vget_low_s16(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+int32x2_t test_vget_low_s32(int32x4_t a) {
+ // CHECK: test_vget_low_s32
+ return vget_low_s32(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+int64x1_t test_vget_low_s64(int64x2_t a) {
+ // CHECK: test_vget_low_s64
+ return vget_low_s64(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+uint8x8_t test_vget_low_u8(uint8x16_t a) {
+ // CHECK: test_vget_low_u8
+ return vget_low_u8(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+uint16x4_t test_vget_low_u16(uint16x8_t a) {
+ // CHECK: test_vget_low_u16
+ return vget_low_u16(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+uint32x2_t test_vget_low_u32(uint32x4_t a) {
+ // CHECK: test_vget_low_u32
+ return vget_low_u32(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+uint64x1_t test_vget_low_u64(uint64x2_t a) {
+ // CHECK: test_vget_low_u64
+ return vget_low_u64(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+poly64x1_t test_vget_low_p64(poly64x2_t a) {
+ // CHECK: test_vget_low_p64
+ return vget_low_p64(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+float16x4_t test_vget_low_f16(float16x8_t a) {
+ // CHECK: test_vget_low_f16
+ return vget_low_f16(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+float32x2_t test_vget_low_f32(float32x4_t a) {
+ // CHECK: test_vget_low_f32
+ return vget_low_f32(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+poly8x8_t test_vget_low_p8(poly8x16_t a) {
+ // CHECK: test_vget_low_p8
+ return vget_low_p8(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+poly16x4_t test_vget_low_p16(poly16x8_t a) {
+ // CHECK: test_vget_low_p16
+ return vget_low_p16(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
+float64x1_t test_vget_low_f64(float64x2_t a) {
+ // CHECK: test_vget_low_f64
+ return vget_low_f64(a);
+ // CHECK: dup d0, {{v[0-9]+}}.d[0]
+}
+
More information about the cfe-commits
mailing list