[llvm] 0ac4f6b - [ARM] MVE vector reduce MLA tests. NFC.
David Green via llvm-commits
llvm-commits at lists.llvm.org
Mon Feb 17 03:54:31 PST 2020
Author: David Green
Date: 2020-02-17T11:54:04Z
New Revision: 0ac4f6b627ec6e9c7f5d137ac425030899dbaf1d
URL: https://github.com/llvm/llvm-project/commit/0ac4f6b627ec6e9c7f5d137ac425030899dbaf1d
DIFF: https://github.com/llvm/llvm-project/commit/0ac4f6b627ec6e9c7f5d137ac425030899dbaf1d.diff
LOG: [ARM] MVE vector reduce MLA tests. NFC.
Added:
llvm/test/CodeGen/Thumb2/mve-vecreduce-mla.ll
Modified:
Removed:
################################################################################
diff --git a/llvm/test/CodeGen/Thumb2/mve-vecreduce-mla.ll b/llvm/test/CodeGen/Thumb2/mve-vecreduce-mla.ll
new file mode 100644
index 000000000000..9b3398ef5126
--- /dev/null
+++ b/llvm/test/CodeGen/Thumb2/mve-vecreduce-mla.ll
@@ -0,0 +1,3035 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=thumbv8.1m.main-arm-none-eabi -mattr=+mve -verify-machineinstrs %s -o - | FileCheck %s --check-prefix=CHECK
+
+define arm_aapcs_vfpcc i32 @add_v4i32_v4i32(<4 x i32> %x, <4 x i32> %y) {
+; CHECK-LABEL: add_v4i32_v4i32:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vaddv.u32 r0, q0
+; CHECK-NEXT: bx lr
+entry:
+ %m = mul <4 x i32> %x, %y
+ %z = call i32 @llvm.experimental.vector.reduce.add.v4i32(<4 x i32> %m)
+ ret i32 %z
+}
+
+define arm_aapcs_vfpcc i64 @add_v4i32_v4i64_zext(<4 x i32> %x, <4 x i32> %y) {
+; CHECK-LABEL: add_v4i32_v4i64_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .vsave {d8, d9}
+; CHECK-NEXT: vpush {d8, d9}
+; CHECK-NEXT: vmov.f32 s8, s4
+; CHECK-NEXT: vmov.f32 s12, s0
+; CHECK-NEXT: vmov.f32 s10, s5
+; CHECK-NEXT: vmov.f32 s14, s1
+; CHECK-NEXT: vmov r0, s8
+; CHECK-NEXT: vmov r1, s12
+; CHECK-NEXT: umull r0, r1, r1, r0
+; CHECK-NEXT: vmov.32 q4[0], r0
+; CHECK-NEXT: vmov r0, s10
+; CHECK-NEXT: vmov.32 q4[1], r1
+; CHECK-NEXT: vmov r1, s14
+; CHECK-NEXT: vmov.f32 s8, s6
+; CHECK-NEXT: vmov.f32 s12, s2
+; CHECK-NEXT: vmov.f32 s10, s7
+; CHECK-NEXT: vmov.f32 s14, s3
+; CHECK-NEXT: umull r0, r1, r1, r0
+; CHECK-NEXT: vmov.32 q4[2], r0
+; CHECK-NEXT: vmov.32 q4[3], r1
+; CHECK-NEXT: vmov r0, s18
+; CHECK-NEXT: vmov r3, s16
+; CHECK-NEXT: vmov r2, s17
+; CHECK-NEXT: adds r0, r0, r3
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: adr r2, .LCPI1_0
+; CHECK-NEXT: vldrw.u32 q1, [r2]
+; CHECK-NEXT: vand q2, q2, q1
+; CHECK-NEXT: vand q0, q3, q1
+; CHECK-NEXT: vmov r2, s8
+; CHECK-NEXT: vmov r3, s0
+; CHECK-NEXT: umlal r0, r1, r3, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: umlal r0, r1, r3, r2
+; CHECK-NEXT: vpop {d8, d9}
+; CHECK-NEXT: bx lr
+; CHECK-NEXT: .p2align 4
+; CHECK-NEXT: @ %bb.1:
+; CHECK-NEXT: .LCPI1_0:
+; CHECK-NEXT: .long 4294967295 @ 0xffffffff
+; CHECK-NEXT: .long 0 @ 0x0
+; CHECK-NEXT: .long 4294967295 @ 0xffffffff
+; CHECK-NEXT: .long 0 @ 0x0
+entry:
+ %xx = zext <4 x i32> %x to <4 x i64>
+ %yy = zext <4 x i32> %y to <4 x i64>
+ %m = mul <4 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v4i64(<4 x i64> %m)
+ ret i64 %z
+}
+
+define arm_aapcs_vfpcc i64 @add_v4i32_v4i64_sext(<4 x i32> %x, <4 x i32> %y) {
+; CHECK-LABEL: add_v4i32_v4i64_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .vsave {d8, d9}
+; CHECK-NEXT: vpush {d8, d9}
+; CHECK-NEXT: vmov.f32 s8, s4
+; CHECK-NEXT: vmov.f32 s12, s0
+; CHECK-NEXT: vmov.f32 s10, s5
+; CHECK-NEXT: vmov.f32 s14, s1
+; CHECK-NEXT: vmov r0, s8
+; CHECK-NEXT: vmov r1, s12
+; CHECK-NEXT: smull r0, r1, r1, r0
+; CHECK-NEXT: vmov.32 q4[0], r0
+; CHECK-NEXT: vmov r0, s10
+; CHECK-NEXT: vmov.32 q4[1], r1
+; CHECK-NEXT: vmov r1, s14
+; CHECK-NEXT: vmov.f32 s8, s6
+; CHECK-NEXT: vmov.f32 s10, s7
+; CHECK-NEXT: vmov.f32 s4, s2
+; CHECK-NEXT: vmov.f32 s6, s3
+; CHECK-NEXT: smull r0, r1, r1, r0
+; CHECK-NEXT: vmov.32 q4[2], r0
+; CHECK-NEXT: vmov.32 q4[3], r1
+; CHECK-NEXT: vmov r0, s18
+; CHECK-NEXT: vmov r3, s16
+; CHECK-NEXT: vmov r2, s17
+; CHECK-NEXT: adds r0, r0, r3
+; CHECK-NEXT: vmov r3, s4
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov r2, s8
+; CHECK-NEXT: smlal r0, r1, r3, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: vmov r3, s6
+; CHECK-NEXT: smlal r0, r1, r3, r2
+; CHECK-NEXT: vpop {d8, d9}
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <4 x i32> %x to <4 x i64>
+ %yy = sext <4 x i32> %y to <4 x i64>
+ %m = mul <4 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v4i64(<4 x i64> %m)
+ ret i64 %z
+}
+
+define arm_aapcs_vfpcc i64 @add_v2i32_v2i64_zext(<2 x i32> %x, <2 x i32> %y) {
+; CHECK-LABEL: add_v2i32_v2i64_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov r0, s4
+; CHECK-NEXT: vmov r1, s0
+; CHECK-NEXT: vmov r2, s6
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: umull r0, r1, r1, r0
+; CHECK-NEXT: umlal r0, r1, r3, r2
+; CHECK-NEXT: bx lr
+entry:
+ %xx = zext <2 x i32> %x to <2 x i64>
+ %yy = zext <2 x i32> %y to <2 x i64>
+ %m = mul <2 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64> %m)
+ ret i64 %z
+}
+
+define arm_aapcs_vfpcc i64 @add_v2i32_v2i64_sext(<2 x i32> %x, <2 x i32> %y) {
+; CHECK-LABEL: add_v2i32_v2i64_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov r0, s4
+; CHECK-NEXT: vmov r1, s0
+; CHECK-NEXT: vmov r2, s6
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: smull r0, r1, r1, r0
+; CHECK-NEXT: smlal r0, r1, r3, r2
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <2 x i32> %x to <2 x i64>
+ %yy = sext <2 x i32> %y to <2 x i64>
+ %m = mul <2 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64> %m)
+ ret i64 %z
+}
+
+define arm_aapcs_vfpcc i32 @add_v8i16_v8i32_zext(<8 x i16> %x, <8 x i16> %y) {
+; CHECK-LABEL: add_v8i16_v8i32_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov.u16 r0, q1[4]
+; CHECK-NEXT: vmov.32 q2[0], r0
+; CHECK-NEXT: vmov.u16 r0, q1[5]
+; CHECK-NEXT: vmov.32 q2[1], r0
+; CHECK-NEXT: vmov.u16 r0, q1[6]
+; CHECK-NEXT: vmov.32 q2[2], r0
+; CHECK-NEXT: vmov.u16 r0, q1[7]
+; CHECK-NEXT: vmov.32 q2[3], r0
+; CHECK-NEXT: vmov.u16 r0, q0[4]
+; CHECK-NEXT: vmov.32 q3[0], r0
+; CHECK-NEXT: vmov.u16 r0, q0[5]
+; CHECK-NEXT: vmov.32 q3[1], r0
+; CHECK-NEXT: vmov.u16 r0, q0[6]
+; CHECK-NEXT: vmov.32 q3[2], r0
+; CHECK-NEXT: vmov.u16 r0, q0[7]
+; CHECK-NEXT: vmov.32 q3[3], r0
+; CHECK-NEXT: vmovlb.u16 q2, q2
+; CHECK-NEXT: vmovlb.u16 q3, q3
+; CHECK-NEXT: vmov.u16 r0, q1[0]
+; CHECK-NEXT: vmul.i32 q2, q3, q2
+; CHECK-NEXT: vmov.32 q3[0], r0
+; CHECK-NEXT: vmov.u16 r0, q1[1]
+; CHECK-NEXT: vmov.32 q3[1], r0
+; CHECK-NEXT: vmov.u16 r0, q1[2]
+; CHECK-NEXT: vmov.32 q3[2], r0
+; CHECK-NEXT: vmov.u16 r0, q1[3]
+; CHECK-NEXT: vmov.32 q3[3], r0
+; CHECK-NEXT: vmov.u16 r0, q0[0]
+; CHECK-NEXT: vmovlb.u16 q1, q3
+; CHECK-NEXT: vmov.32 q3[0], r0
+; CHECK-NEXT: vmov.u16 r0, q0[1]
+; CHECK-NEXT: vmov.32 q3[1], r0
+; CHECK-NEXT: vmov.u16 r0, q0[2]
+; CHECK-NEXT: vmov.32 q3[2], r0
+; CHECK-NEXT: vmov.u16 r0, q0[3]
+; CHECK-NEXT: vmov.32 q3[3], r0
+; CHECK-NEXT: vmovlb.u16 q0, q3
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vadd.i32 q0, q0, q2
+; CHECK-NEXT: vaddv.u32 r0, q0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = zext <8 x i16> %x to <8 x i32>
+ %yy = zext <8 x i16> %y to <8 x i32>
+ %m = mul <8 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v8i32(<8 x i32> %m)
+ ret i32 %z
+}
+
+define arm_aapcs_vfpcc i32 @add_v8i16_v8i32_sext(<8 x i16> %x, <8 x i16> %y) {
+; CHECK-LABEL: add_v8i16_v8i32_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov.u16 r0, q1[4]
+; CHECK-NEXT: vmov.32 q2[0], r0
+; CHECK-NEXT: vmov.u16 r0, q1[5]
+; CHECK-NEXT: vmov.32 q2[1], r0
+; CHECK-NEXT: vmov.u16 r0, q1[6]
+; CHECK-NEXT: vmov.32 q2[2], r0
+; CHECK-NEXT: vmov.u16 r0, q1[7]
+; CHECK-NEXT: vmov.32 q2[3], r0
+; CHECK-NEXT: vmov.u16 r0, q0[4]
+; CHECK-NEXT: vmov.32 q3[0], r0
+; CHECK-NEXT: vmov.u16 r0, q0[5]
+; CHECK-NEXT: vmov.32 q3[1], r0
+; CHECK-NEXT: vmov.u16 r0, q0[6]
+; CHECK-NEXT: vmov.32 q3[2], r0
+; CHECK-NEXT: vmov.u16 r0, q0[7]
+; CHECK-NEXT: vmov.32 q3[3], r0
+; CHECK-NEXT: vmovlb.s16 q2, q2
+; CHECK-NEXT: vmovlb.s16 q3, q3
+; CHECK-NEXT: vmov.u16 r0, q1[0]
+; CHECK-NEXT: vmul.i32 q2, q3, q2
+; CHECK-NEXT: vmov.32 q3[0], r0
+; CHECK-NEXT: vmov.u16 r0, q1[1]
+; CHECK-NEXT: vmov.32 q3[1], r0
+; CHECK-NEXT: vmov.u16 r0, q1[2]
+; CHECK-NEXT: vmov.32 q3[2], r0
+; CHECK-NEXT: vmov.u16 r0, q1[3]
+; CHECK-NEXT: vmov.32 q3[3], r0
+; CHECK-NEXT: vmov.u16 r0, q0[0]
+; CHECK-NEXT: vmovlb.s16 q1, q3
+; CHECK-NEXT: vmov.32 q3[0], r0
+; CHECK-NEXT: vmov.u16 r0, q0[1]
+; CHECK-NEXT: vmov.32 q3[1], r0
+; CHECK-NEXT: vmov.u16 r0, q0[2]
+; CHECK-NEXT: vmov.32 q3[2], r0
+; CHECK-NEXT: vmov.u16 r0, q0[3]
+; CHECK-NEXT: vmov.32 q3[3], r0
+; CHECK-NEXT: vmovlb.s16 q0, q3
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vadd.i32 q0, q0, q2
+; CHECK-NEXT: vaddv.u32 r0, q0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <8 x i16> %x to <8 x i32>
+ %yy = sext <8 x i16> %y to <8 x i32>
+ %m = mul <8 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v8i32(<8 x i32> %m)
+ ret i32 %z
+}
+
+define arm_aapcs_vfpcc i32 @add_v4i16_v4i32_zext(<4 x i16> %x, <4 x i16> %y) {
+; CHECK-LABEL: add_v4i16_v4i32_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmovlb.u16 q1, q1
+; CHECK-NEXT: vmovlb.u16 q0, q0
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vaddv.u32 r0, q0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = zext <4 x i16> %x to <4 x i32>
+ %yy = zext <4 x i16> %y to <4 x i32>
+ %m = mul <4 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v4i32(<4 x i32> %m)
+ ret i32 %z
+}
+
+define arm_aapcs_vfpcc i32 @add_v4i16_v4i32_sext(<4 x i16> %x, <4 x i16> %y) {
+; CHECK-LABEL: add_v4i16_v4i32_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmovlb.s16 q1, q1
+; CHECK-NEXT: vmovlb.s16 q0, q0
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vaddv.u32 r0, q0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <4 x i16> %x to <4 x i32>
+ %yy = sext <4 x i16> %y to <4 x i32>
+ %m = mul <4 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v4i32(<4 x i32> %m)
+ ret i32 %z
+}
+
+define arm_aapcs_vfpcc zeroext i16 @add_v8i16_v8i16(<8 x i16> %x, <8 x i16> %y) {
+; CHECK-LABEL: add_v8i16_v8i16:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmul.i16 q0, q0, q1
+; CHECK-NEXT: vaddv.u16 r0, q0
+; CHECK-NEXT: uxth r0, r0
+; CHECK-NEXT: bx lr
+entry:
+ %m = mul <8 x i16> %x, %y
+ %z = call i16 @llvm.experimental.vector.reduce.add.v8i16(<8 x i16> %m)
+ ret i16 %z
+}
+
+define arm_aapcs_vfpcc i64 @add_v8i16_v8i64_zext(<8 x i16> %x, <8 x i16> %y) {
+; CHECK-LABEL: add_v8i16_v8i64_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-NEXT: vmov.u16 r0, q1[0]
+; CHECK-NEXT: vmov.u16 r1, q0[0]
+; CHECK-NEXT: vmov.32 q3[0], r0
+; CHECK-NEXT: vmov.u16 r0, q1[1]
+; CHECK-NEXT: vmov.32 q3[2], r0
+; CHECK-NEXT: adr r0, .LCPI10_0
+; CHECK-NEXT: vldrw.u32 q2, [r0]
+; CHECK-NEXT: vmov.32 q4[0], r1
+; CHECK-NEXT: vmov.u16 r1, q0[1]
+; CHECK-NEXT: vmov.32 q4[2], r1
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r0, s12
+; CHECK-NEXT: vmov r1, s16
+; CHECK-NEXT: umull r0, r1, r1, r0
+; CHECK-NEXT: vmov.32 q5[0], r0
+; CHECK-NEXT: vmov r0, s14
+; CHECK-NEXT: vmov.32 q5[1], r1
+; CHECK-NEXT: vmov r1, s18
+; CHECK-NEXT: umull r0, r1, r1, r0
+; CHECK-NEXT: vmov.32 q5[2], r0
+; CHECK-NEXT: vmov.32 q5[3], r1
+; CHECK-NEXT: vmov r0, s22
+; CHECK-NEXT: vmov r3, s20
+; CHECK-NEXT: vmov r2, s21
+; CHECK-NEXT: adds.w r12, r3, r0
+; CHECK-NEXT: vmov.u16 r3, q0[2]
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov.u16 r2, q1[2]
+; CHECK-NEXT: vmov.32 q3[0], r2
+; CHECK-NEXT: vmov.u16 r2, q1[3]
+; CHECK-NEXT: vmov.32 q4[0], r3
+; CHECK-NEXT: vmov.u16 r3, q0[3]
+; CHECK-NEXT: vmov.32 q3[2], r2
+; CHECK-NEXT: vmov.32 q4[2], r3
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r2, s12
+; CHECK-NEXT: vmov r3, s16
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[0], r2
+; CHECK-NEXT: vmov r2, s14
+; CHECK-NEXT: vmov.32 q5[1], r3
+; CHECK-NEXT: vmov r3, s18
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[2], r2
+; CHECK-NEXT: vmov.32 q5[3], r3
+; CHECK-NEXT: vmov r0, s20
+; CHECK-NEXT: vmov r2, s21
+; CHECK-NEXT: adds.w r0, r0, r12
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov r2, s22
+; CHECK-NEXT: adds.w r12, r0, r2
+; CHECK-NEXT: vmov.u16 r2, q1[4]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.u16 r3, q0[4]
+; CHECK-NEXT: vmov.32 q3[0], r2
+; CHECK-NEXT: vmov.u16 r2, q1[5]
+; CHECK-NEXT: vmov.32 q4[0], r3
+; CHECK-NEXT: vmov.u16 r3, q0[5]
+; CHECK-NEXT: vmov.32 q3[2], r2
+; CHECK-NEXT: vmov.32 q4[2], r3
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r2, s12
+; CHECK-NEXT: vmov r3, s16
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[0], r2
+; CHECK-NEXT: vmov r2, s14
+; CHECK-NEXT: vmov.32 q5[1], r3
+; CHECK-NEXT: vmov r3, s18
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[2], r2
+; CHECK-NEXT: vmov.32 q5[3], r3
+; CHECK-NEXT: vmov r0, s20
+; CHECK-NEXT: vmov r2, s21
+; CHECK-NEXT: adds.w r0, r0, r12
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov r2, s22
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: vmov.u16 r2, q1[6]
+; CHECK-NEXT: vmov.32 q3[0], r2
+; CHECK-NEXT: vmov.u16 r2, q1[7]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.32 q3[2], r2
+; CHECK-NEXT: vmov.u16 r3, q0[6]
+; CHECK-NEXT: vand q1, q3, q2
+; CHECK-NEXT: vmov.32 q3[0], r3
+; CHECK-NEXT: vmov.u16 r3, q0[7]
+; CHECK-NEXT: vmov.32 q3[2], r3
+; CHECK-NEXT: vmov r2, s4
+; CHECK-NEXT: vand q0, q3, q2
+; CHECK-NEXT: vmov r3, s0
+; CHECK-NEXT: umlal r0, r1, r3, r2
+; CHECK-NEXT: vmov r2, s6
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: umlal r0, r1, r3, r2
+; CHECK-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-NEXT: bx lr
+; CHECK-NEXT: .p2align 4
+; CHECK-NEXT: @ %bb.1:
+; CHECK-NEXT: .LCPI10_0:
+; CHECK-NEXT: .long 65535 @ 0xffff
+; CHECK-NEXT: .long 0 @ 0x0
+; CHECK-NEXT: .long 65535 @ 0xffff
+; CHECK-NEXT: .long 0 @ 0x0
+entry:
+ %xx = zext <8 x i16> %x to <8 x i64>
+ %yy = zext <8 x i16> %y to <8 x i64>
+ %m = mul <8 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v8i64(<8 x i64> %m)
+ ret i64 %z
+}
+
+define arm_aapcs_vfpcc i64 @add_v8i16_v8i64_sext(<8 x i16> %x, <8 x i16> %y) {
+; CHECK-LABEL: add_v8i16_v8i64_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov.u16 r0, q1[0]
+; CHECK-NEXT: vmov.u16 r1, q0[0]
+; CHECK-NEXT: sxth r0, r0
+; CHECK-NEXT: sxth r1, r1
+; CHECK-NEXT: smull r0, r1, r1, r0
+; CHECK-NEXT: vmov.32 q2[0], r0
+; CHECK-NEXT: vmov.u16 r0, q1[1]
+; CHECK-NEXT: vmov.32 q2[1], r1
+; CHECK-NEXT: vmov.u16 r1, q0[1]
+; CHECK-NEXT: sxth r0, r0
+; CHECK-NEXT: sxth r1, r1
+; CHECK-NEXT: smull r0, r1, r1, r0
+; CHECK-NEXT: vmov.32 q2[2], r0
+; CHECK-NEXT: vmov.32 q2[3], r1
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: vmov r3, s8
+; CHECK-NEXT: vmov r0, s9
+; CHECK-NEXT: adds r2, r2, r3
+; CHECK-NEXT: vmov.u16 r3, q0[2]
+; CHECK-NEXT: adc.w r12, r0, r1
+; CHECK-NEXT: vmov.u16 r1, q1[2]
+; CHECK-NEXT: sxth r1, r1
+; CHECK-NEXT: sxth r3, r3
+; CHECK-NEXT: smull r1, r3, r3, r1
+; CHECK-NEXT: vmov.32 q2[0], r1
+; CHECK-NEXT: vmov.u16 r1, q1[3]
+; CHECK-NEXT: vmov.32 q2[1], r3
+; CHECK-NEXT: vmov.u16 r3, q0[3]
+; CHECK-NEXT: sxth r1, r1
+; CHECK-NEXT: sxth r3, r3
+; CHECK-NEXT: smull r1, r3, r3, r1
+; CHECK-NEXT: vmov.32 q2[2], r1
+; CHECK-NEXT: vmov.32 q2[3], r3
+; CHECK-NEXT: vmov r0, s8
+; CHECK-NEXT: vmov r1, s9
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adc.w r1, r1, r12
+; CHECK-NEXT: adds.w r12, r0, r2
+; CHECK-NEXT: vmov.u16 r2, q1[4]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.u16 r3, q0[4]
+; CHECK-NEXT: sxth r2, r2
+; CHECK-NEXT: sxth r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u16 r2, q1[5]
+; CHECK-NEXT: vmov.32 q2[1], r3
+; CHECK-NEXT: vmov.u16 r3, q0[5]
+; CHECK-NEXT: sxth r2, r2
+; CHECK-NEXT: sxth r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r3
+; CHECK-NEXT: vmov r0, s8
+; CHECK-NEXT: vmov r2, s9
+; CHECK-NEXT: adds.w r0, r0, r12
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: vmov.u16 r2, q1[6]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.u16 r3, q0[6]
+; CHECK-NEXT: sxth r2, r2
+; CHECK-NEXT: sxth r3, r3
+; CHECK-NEXT: smlal r0, r1, r3, r2
+; CHECK-NEXT: vmov.u16 r2, q1[7]
+; CHECK-NEXT: vmov.u16 r3, q0[7]
+; CHECK-NEXT: sxth r2, r2
+; CHECK-NEXT: sxth r3, r3
+; CHECK-NEXT: smlal r0, r1, r3, r2
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <8 x i16> %x to <8 x i64>
+ %yy = sext <8 x i16> %y to <8 x i64>
+ %m = mul <8 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v8i64(<8 x i64> %m)
+ ret i64 %z
+}
+
+define arm_aapcs_vfpcc i64 @add_v2i16_v2i64_zext(<2 x i16> %x, <2 x i16> %y) {
+; CHECK-LABEL: add_v2i16_v2i64_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: adr r0, .LCPI12_0
+; CHECK-NEXT: vldrw.u32 q2, [r0]
+; CHECK-NEXT: vand q1, q1, q2
+; CHECK-NEXT: vand q0, q0, q2
+; CHECK-NEXT: vmov r0, s4
+; CHECK-NEXT: vmov r1, s0
+; CHECK-NEXT: vmov r2, s6
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: umull r0, r1, r1, r0
+; CHECK-NEXT: umlal r0, r1, r3, r2
+; CHECK-NEXT: bx lr
+; CHECK-NEXT: .p2align 4
+; CHECK-NEXT: @ %bb.1:
+; CHECK-NEXT: .LCPI12_0:
+; CHECK-NEXT: .long 65535 @ 0xffff
+; CHECK-NEXT: .long 0 @ 0x0
+; CHECK-NEXT: .long 65535 @ 0xffff
+; CHECK-NEXT: .long 0 @ 0x0
+entry:
+ %xx = zext <2 x i16> %x to <2 x i64>
+ %yy = zext <2 x i16> %y to <2 x i64>
+ %m = mul <2 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64> %m)
+ ret i64 %z
+}
+
+define arm_aapcs_vfpcc i64 @add_v2i16_v2i64_sext(<2 x i16> %x, <2 x i16> %y) {
+; CHECK-LABEL: add_v2i16_v2i64_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov r0, s4
+; CHECK-NEXT: vmov r1, s0
+; CHECK-NEXT: vmov r2, s6
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: sxth r0, r0
+; CHECK-NEXT: sxth r1, r1
+; CHECK-NEXT: smull r0, r1, r1, r0
+; CHECK-NEXT: sxth r2, r2
+; CHECK-NEXT: sxth r3, r3
+; CHECK-NEXT: smlal r0, r1, r3, r2
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <2 x i16> %x to <2 x i64>
+ %yy = sext <2 x i16> %y to <2 x i64>
+ %m = mul <2 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64> %m)
+ ret i64 %z
+}
+
+define arm_aapcs_vfpcc i32 @add_v16i8_v16i32_zext(<16 x i8> %x, <16 x i8> %y) {
+; CHECK-LABEL: add_v16i8_v16i32_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-NEXT: vmov.u8 r0, q1[12]
+; CHECK-NEXT: vmov.i32 q2, #0xff
+; CHECK-NEXT: vmov.32 q3[0], r0
+; CHECK-NEXT: vmov.u8 r0, q1[13]
+; CHECK-NEXT: vmov.32 q3[1], r0
+; CHECK-NEXT: vmov.u8 r0, q1[14]
+; CHECK-NEXT: vmov.32 q3[2], r0
+; CHECK-NEXT: vmov.u8 r0, q1[15]
+; CHECK-NEXT: vmov.32 q3[3], r0
+; CHECK-NEXT: vmov.u8 r0, q0[12]
+; CHECK-NEXT: vmov.32 q4[0], r0
+; CHECK-NEXT: vmov.u8 r0, q0[13]
+; CHECK-NEXT: vmov.32 q4[1], r0
+; CHECK-NEXT: vmov.u8 r0, q0[14]
+; CHECK-NEXT: vmov.32 q4[2], r0
+; CHECK-NEXT: vmov.u8 r0, q0[15]
+; CHECK-NEXT: vmov.32 q4[3], r0
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov.u8 r0, q1[4]
+; CHECK-NEXT: vmul.i32 q3, q4, q3
+; CHECK-NEXT: vmov.32 q4[0], r0
+; CHECK-NEXT: vmov.u8 r0, q1[5]
+; CHECK-NEXT: vmov.32 q4[1], r0
+; CHECK-NEXT: vmov.u8 r0, q1[6]
+; CHECK-NEXT: vmov.32 q4[2], r0
+; CHECK-NEXT: vmov.u8 r0, q1[7]
+; CHECK-NEXT: vmov.32 q4[3], r0
+; CHECK-NEXT: vmov.u8 r0, q0[4]
+; CHECK-NEXT: vmov.32 q5[0], r0
+; CHECK-NEXT: vmov.u8 r0, q0[5]
+; CHECK-NEXT: vmov.32 q5[1], r0
+; CHECK-NEXT: vmov.u8 r0, q0[6]
+; CHECK-NEXT: vmov.32 q5[2], r0
+; CHECK-NEXT: vmov.u8 r0, q0[7]
+; CHECK-NEXT: vmov.32 q5[3], r0
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vand q5, q5, q2
+; CHECK-NEXT: vmov.u8 r0, q1[8]
+; CHECK-NEXT: vmul.i32 q4, q5, q4
+; CHECK-NEXT: vadd.i32 q3, q4, q3
+; CHECK-NEXT: vmov.32 q4[0], r0
+; CHECK-NEXT: vmov.u8 r0, q1[9]
+; CHECK-NEXT: vmov.32 q4[1], r0
+; CHECK-NEXT: vmov.u8 r0, q1[10]
+; CHECK-NEXT: vmov.32 q4[2], r0
+; CHECK-NEXT: vmov.u8 r0, q1[11]
+; CHECK-NEXT: vmov.32 q4[3], r0
+; CHECK-NEXT: vmov.u8 r0, q0[8]
+; CHECK-NEXT: vmov.32 q5[0], r0
+; CHECK-NEXT: vmov.u8 r0, q0[9]
+; CHECK-NEXT: vmov.32 q5[1], r0
+; CHECK-NEXT: vmov.u8 r0, q0[10]
+; CHECK-NEXT: vmov.32 q5[2], r0
+; CHECK-NEXT: vmov.u8 r0, q0[11]
+; CHECK-NEXT: vmov.32 q5[3], r0
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vand q5, q5, q2
+; CHECK-NEXT: vmov.u8 r0, q1[0]
+; CHECK-NEXT: vmul.i32 q4, q5, q4
+; CHECK-NEXT: vmov.32 q5[0], r0
+; CHECK-NEXT: vmov.u8 r0, q1[1]
+; CHECK-NEXT: vmov.32 q5[1], r0
+; CHECK-NEXT: vmov.u8 r0, q1[2]
+; CHECK-NEXT: vmov.32 q5[2], r0
+; CHECK-NEXT: vmov.u8 r0, q1[3]
+; CHECK-NEXT: vmov.32 q5[3], r0
+; CHECK-NEXT: vmov.u8 r0, q0[0]
+; CHECK-NEXT: vand q1, q5, q2
+; CHECK-NEXT: vmov.32 q5[0], r0
+; CHECK-NEXT: vmov.u8 r0, q0[1]
+; CHECK-NEXT: vmov.32 q5[1], r0
+; CHECK-NEXT: vmov.u8 r0, q0[2]
+; CHECK-NEXT: vmov.32 q5[2], r0
+; CHECK-NEXT: vmov.u8 r0, q0[3]
+; CHECK-NEXT: vmov.32 q5[3], r0
+; CHECK-NEXT: vand q0, q5, q2
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vadd.i32 q0, q0, q4
+; CHECK-NEXT: vadd.i32 q0, q0, q3
+; CHECK-NEXT: vaddv.u32 r0, q0
+; CHECK-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-NEXT: bx lr
+entry:
+ %xx = zext <16 x i8> %x to <16 x i32>
+ %yy = zext <16 x i8> %y to <16 x i32>
+ %m = mul <16 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v16i32(<16 x i32> %m)
+ ret i32 %z
+}
+
+define arm_aapcs_vfpcc i32 @add_v16i8_v16i32_sext(<16 x i8> %x, <16 x i8> %y) {
+; CHECK-LABEL: add_v16i8_v16i32_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .vsave {d8, d9}
+; CHECK-NEXT: vpush {d8, d9}
+; CHECK-NEXT: vmov.u8 r0, q1[12]
+; CHECK-NEXT: vmov.32 q2[0], r0
+; CHECK-NEXT: vmov.u8 r0, q1[13]
+; CHECK-NEXT: vmov.32 q2[1], r0
+; CHECK-NEXT: vmov.u8 r0, q1[14]
+; CHECK-NEXT: vmov.32 q2[2], r0
+; CHECK-NEXT: vmov.u8 r0, q1[15]
+; CHECK-NEXT: vmov.32 q2[3], r0
+; CHECK-NEXT: vmov.u8 r0, q0[12]
+; CHECK-NEXT: vmov.32 q3[0], r0
+; CHECK-NEXT: vmov.u8 r0, q0[13]
+; CHECK-NEXT: vmov.32 q3[1], r0
+; CHECK-NEXT: vmov.u8 r0, q0[14]
+; CHECK-NEXT: vmov.32 q3[2], r0
+; CHECK-NEXT: vmov.u8 r0, q0[15]
+; CHECK-NEXT: vmov.32 q3[3], r0
+; CHECK-NEXT: vmovlb.s8 q2, q2
+; CHECK-NEXT: vmovlb.s8 q3, q3
+; CHECK-NEXT: vmovlb.s16 q2, q2
+; CHECK-NEXT: vmovlb.s16 q3, q3
+; CHECK-NEXT: vmov.u8 r0, q1[4]
+; CHECK-NEXT: vmul.i32 q2, q3, q2
+; CHECK-NEXT: vmov.32 q3[0], r0
+; CHECK-NEXT: vmov.u8 r0, q1[5]
+; CHECK-NEXT: vmov.32 q3[1], r0
+; CHECK-NEXT: vmov.u8 r0, q1[6]
+; CHECK-NEXT: vmov.32 q3[2], r0
+; CHECK-NEXT: vmov.u8 r0, q1[7]
+; CHECK-NEXT: vmov.32 q3[3], r0
+; CHECK-NEXT: vmov.u8 r0, q0[4]
+; CHECK-NEXT: vmov.32 q4[0], r0
+; CHECK-NEXT: vmov.u8 r0, q0[5]
+; CHECK-NEXT: vmov.32 q4[1], r0
+; CHECK-NEXT: vmov.u8 r0, q0[6]
+; CHECK-NEXT: vmov.32 q4[2], r0
+; CHECK-NEXT: vmov.u8 r0, q0[7]
+; CHECK-NEXT: vmov.32 q4[3], r0
+; CHECK-NEXT: vmovlb.s8 q3, q3
+; CHECK-NEXT: vmovlb.s8 q4, q4
+; CHECK-NEXT: vmovlb.s16 q3, q3
+; CHECK-NEXT: vmovlb.s16 q4, q4
+; CHECK-NEXT: vmov.u8 r0, q1[8]
+; CHECK-NEXT: vmul.i32 q3, q4, q3
+; CHECK-NEXT: vadd.i32 q2, q3, q2
+; CHECK-NEXT: vmov.32 q3[0], r0
+; CHECK-NEXT: vmov.u8 r0, q1[9]
+; CHECK-NEXT: vmov.32 q3[1], r0
+; CHECK-NEXT: vmov.u8 r0, q1[10]
+; CHECK-NEXT: vmov.32 q3[2], r0
+; CHECK-NEXT: vmov.u8 r0, q1[11]
+; CHECK-NEXT: vmov.32 q3[3], r0
+; CHECK-NEXT: vmov.u8 r0, q0[8]
+; CHECK-NEXT: vmov.32 q4[0], r0
+; CHECK-NEXT: vmov.u8 r0, q0[9]
+; CHECK-NEXT: vmov.32 q4[1], r0
+; CHECK-NEXT: vmov.u8 r0, q0[10]
+; CHECK-NEXT: vmov.32 q4[2], r0
+; CHECK-NEXT: vmov.u8 r0, q0[11]
+; CHECK-NEXT: vmov.32 q4[3], r0
+; CHECK-NEXT: vmovlb.s8 q3, q3
+; CHECK-NEXT: vmovlb.s8 q4, q4
+; CHECK-NEXT: vmovlb.s16 q3, q3
+; CHECK-NEXT: vmovlb.s16 q4, q4
+; CHECK-NEXT: vmov.u8 r0, q1[0]
+; CHECK-NEXT: vmul.i32 q3, q4, q3
+; CHECK-NEXT: vmov.32 q4[0], r0
+; CHECK-NEXT: vmov.u8 r0, q1[1]
+; CHECK-NEXT: vmov.32 q4[1], r0
+; CHECK-NEXT: vmov.u8 r0, q1[2]
+; CHECK-NEXT: vmov.32 q4[2], r0
+; CHECK-NEXT: vmov.u8 r0, q1[3]
+; CHECK-NEXT: vmov.32 q4[3], r0
+; CHECK-NEXT: vmov.u8 r0, q0[0]
+; CHECK-NEXT: vmovlb.s8 q1, q4
+; CHECK-NEXT: vmov.32 q4[0], r0
+; CHECK-NEXT: vmov.u8 r0, q0[1]
+; CHECK-NEXT: vmovlb.s16 q1, q1
+; CHECK-NEXT: vmov.32 q4[1], r0
+; CHECK-NEXT: vmov.u8 r0, q0[2]
+; CHECK-NEXT: vmov.32 q4[2], r0
+; CHECK-NEXT: vmov.u8 r0, q0[3]
+; CHECK-NEXT: vmov.32 q4[3], r0
+; CHECK-NEXT: vmovlb.s8 q0, q4
+; CHECK-NEXT: vmovlb.s16 q0, q0
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vadd.i32 q0, q0, q3
+; CHECK-NEXT: vadd.i32 q0, q0, q2
+; CHECK-NEXT: vaddv.u32 r0, q0
+; CHECK-NEXT: vpop {d8, d9}
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <16 x i8> %x to <16 x i32>
+ %yy = sext <16 x i8> %y to <16 x i32>
+ %m = mul <16 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v16i32(<16 x i32> %m)
+ ret i32 %z
+}
+
+define arm_aapcs_vfpcc i32 @add_v4i8_v4i32_zext(<4 x i8> %x, <4 x i8> %y) {
+; CHECK-LABEL: add_v4i8_v4i32_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov.i32 q2, #0xff
+; CHECK-NEXT: vand q1, q1, q2
+; CHECK-NEXT: vand q0, q0, q2
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vaddv.u32 r0, q0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = zext <4 x i8> %x to <4 x i32>
+ %yy = zext <4 x i8> %y to <4 x i32>
+ %m = mul <4 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v4i32(<4 x i32> %m)
+ ret i32 %z
+}
+
+define arm_aapcs_vfpcc i32 @add_v4i8_v4i32_sext(<4 x i8> %x, <4 x i8> %y) {
+; CHECK-LABEL: add_v4i8_v4i32_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmovlb.s8 q1, q1
+; CHECK-NEXT: vmovlb.s8 q0, q0
+; CHECK-NEXT: vmovlb.s16 q1, q1
+; CHECK-NEXT: vmovlb.s16 q0, q0
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vaddv.u32 r0, q0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <4 x i8> %x to <4 x i32>
+ %yy = sext <4 x i8> %y to <4 x i32>
+ %m = mul <4 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v4i32(<4 x i32> %m)
+ ret i32 %z
+}
+
+define arm_aapcs_vfpcc zeroext i16 @add_v16i8_v16i16_zext(<16 x i8> %x, <16 x i8> %y) {
+; CHECK-LABEL: add_v16i8_v16i16_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov.u8 r0, q1[8]
+; CHECK-NEXT: vmov.16 q2[0], r0
+; CHECK-NEXT: vmov.u8 r0, q1[9]
+; CHECK-NEXT: vmov.16 q2[1], r0
+; CHECK-NEXT: vmov.u8 r0, q1[10]
+; CHECK-NEXT: vmov.16 q2[2], r0
+; CHECK-NEXT: vmov.u8 r0, q1[11]
+; CHECK-NEXT: vmov.16 q2[3], r0
+; CHECK-NEXT: vmov.u8 r0, q1[12]
+; CHECK-NEXT: vmov.16 q2[4], r0
+; CHECK-NEXT: vmov.u8 r0, q1[13]
+; CHECK-NEXT: vmov.16 q2[5], r0
+; CHECK-NEXT: vmov.u8 r0, q1[14]
+; CHECK-NEXT: vmov.16 q2[6], r0
+; CHECK-NEXT: vmov.u8 r0, q1[15]
+; CHECK-NEXT: vmov.16 q2[7], r0
+; CHECK-NEXT: vmov.u8 r0, q0[8]
+; CHECK-NEXT: vmov.16 q3[0], r0
+; CHECK-NEXT: vmov.u8 r0, q0[9]
+; CHECK-NEXT: vmov.16 q3[1], r0
+; CHECK-NEXT: vmov.u8 r0, q0[10]
+; CHECK-NEXT: vmov.16 q3[2], r0
+; CHECK-NEXT: vmov.u8 r0, q0[11]
+; CHECK-NEXT: vmov.16 q3[3], r0
+; CHECK-NEXT: vmov.u8 r0, q0[12]
+; CHECK-NEXT: vmov.16 q3[4], r0
+; CHECK-NEXT: vmov.u8 r0, q0[13]
+; CHECK-NEXT: vmov.16 q3[5], r0
+; CHECK-NEXT: vmov.u8 r0, q0[14]
+; CHECK-NEXT: vmov.16 q3[6], r0
+; CHECK-NEXT: vmov.u8 r0, q0[15]
+; CHECK-NEXT: vmov.16 q3[7], r0
+; CHECK-NEXT: vmovlb.u8 q2, q2
+; CHECK-NEXT: vmovlb.u8 q3, q3
+; CHECK-NEXT: vmov.u8 r0, q1[0]
+; CHECK-NEXT: vmul.i16 q2, q3, q2
+; CHECK-NEXT: vmov.16 q3[0], r0
+; CHECK-NEXT: vmov.u8 r0, q1[1]
+; CHECK-NEXT: vmov.16 q3[1], r0
+; CHECK-NEXT: vmov.u8 r0, q1[2]
+; CHECK-NEXT: vmov.16 q3[2], r0
+; CHECK-NEXT: vmov.u8 r0, q1[3]
+; CHECK-NEXT: vmov.16 q3[3], r0
+; CHECK-NEXT: vmov.u8 r0, q1[4]
+; CHECK-NEXT: vmov.16 q3[4], r0
+; CHECK-NEXT: vmov.u8 r0, q1[5]
+; CHECK-NEXT: vmov.16 q3[5], r0
+; CHECK-NEXT: vmov.u8 r0, q1[6]
+; CHECK-NEXT: vmov.16 q3[6], r0
+; CHECK-NEXT: vmov.u8 r0, q1[7]
+; CHECK-NEXT: vmov.16 q3[7], r0
+; CHECK-NEXT: vmov.u8 r0, q0[0]
+; CHECK-NEXT: vmovlb.u8 q1, q3
+; CHECK-NEXT: vmov.16 q3[0], r0
+; CHECK-NEXT: vmov.u8 r0, q0[1]
+; CHECK-NEXT: vmov.16 q3[1], r0
+; CHECK-NEXT: vmov.u8 r0, q0[2]
+; CHECK-NEXT: vmov.16 q3[2], r0
+; CHECK-NEXT: vmov.u8 r0, q0[3]
+; CHECK-NEXT: vmov.16 q3[3], r0
+; CHECK-NEXT: vmov.u8 r0, q0[4]
+; CHECK-NEXT: vmov.16 q3[4], r0
+; CHECK-NEXT: vmov.u8 r0, q0[5]
+; CHECK-NEXT: vmov.16 q3[5], r0
+; CHECK-NEXT: vmov.u8 r0, q0[6]
+; CHECK-NEXT: vmov.16 q3[6], r0
+; CHECK-NEXT: vmov.u8 r0, q0[7]
+; CHECK-NEXT: vmov.16 q3[7], r0
+; CHECK-NEXT: vmovlb.u8 q0, q3
+; CHECK-NEXT: vmul.i16 q0, q0, q1
+; CHECK-NEXT: vadd.i16 q0, q0, q2
+; CHECK-NEXT: vaddv.u16 r0, q0
+; CHECK-NEXT: uxth r0, r0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = zext <16 x i8> %x to <16 x i16>
+ %yy = zext <16 x i8> %y to <16 x i16>
+ %m = mul <16 x i16> %xx, %yy
+ %z = call i16 @llvm.experimental.vector.reduce.add.v16i16(<16 x i16> %m)
+ ret i16 %z
+}
+
+define arm_aapcs_vfpcc signext i16 @add_v16i8_v16i16_sext(<16 x i8> %x, <16 x i8> %y) {
+; CHECK-LABEL: add_v16i8_v16i16_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov.u8 r0, q1[8]
+; CHECK-NEXT: vmov.16 q2[0], r0
+; CHECK-NEXT: vmov.u8 r0, q1[9]
+; CHECK-NEXT: vmov.16 q2[1], r0
+; CHECK-NEXT: vmov.u8 r0, q1[10]
+; CHECK-NEXT: vmov.16 q2[2], r0
+; CHECK-NEXT: vmov.u8 r0, q1[11]
+; CHECK-NEXT: vmov.16 q2[3], r0
+; CHECK-NEXT: vmov.u8 r0, q1[12]
+; CHECK-NEXT: vmov.16 q2[4], r0
+; CHECK-NEXT: vmov.u8 r0, q1[13]
+; CHECK-NEXT: vmov.16 q2[5], r0
+; CHECK-NEXT: vmov.u8 r0, q1[14]
+; CHECK-NEXT: vmov.16 q2[6], r0
+; CHECK-NEXT: vmov.u8 r0, q1[15]
+; CHECK-NEXT: vmov.16 q2[7], r0
+; CHECK-NEXT: vmov.u8 r0, q0[8]
+; CHECK-NEXT: vmov.16 q3[0], r0
+; CHECK-NEXT: vmov.u8 r0, q0[9]
+; CHECK-NEXT: vmov.16 q3[1], r0
+; CHECK-NEXT: vmov.u8 r0, q0[10]
+; CHECK-NEXT: vmov.16 q3[2], r0
+; CHECK-NEXT: vmov.u8 r0, q0[11]
+; CHECK-NEXT: vmov.16 q3[3], r0
+; CHECK-NEXT: vmov.u8 r0, q0[12]
+; CHECK-NEXT: vmov.16 q3[4], r0
+; CHECK-NEXT: vmov.u8 r0, q0[13]
+; CHECK-NEXT: vmov.16 q3[5], r0
+; CHECK-NEXT: vmov.u8 r0, q0[14]
+; CHECK-NEXT: vmov.16 q3[6], r0
+; CHECK-NEXT: vmov.u8 r0, q0[15]
+; CHECK-NEXT: vmov.16 q3[7], r0
+; CHECK-NEXT: vmovlb.s8 q2, q2
+; CHECK-NEXT: vmovlb.s8 q3, q3
+; CHECK-NEXT: vmov.u8 r0, q1[0]
+; CHECK-NEXT: vmul.i16 q2, q3, q2
+; CHECK-NEXT: vmov.16 q3[0], r0
+; CHECK-NEXT: vmov.u8 r0, q1[1]
+; CHECK-NEXT: vmov.16 q3[1], r0
+; CHECK-NEXT: vmov.u8 r0, q1[2]
+; CHECK-NEXT: vmov.16 q3[2], r0
+; CHECK-NEXT: vmov.u8 r0, q1[3]
+; CHECK-NEXT: vmov.16 q3[3], r0
+; CHECK-NEXT: vmov.u8 r0, q1[4]
+; CHECK-NEXT: vmov.16 q3[4], r0
+; CHECK-NEXT: vmov.u8 r0, q1[5]
+; CHECK-NEXT: vmov.16 q3[5], r0
+; CHECK-NEXT: vmov.u8 r0, q1[6]
+; CHECK-NEXT: vmov.16 q3[6], r0
+; CHECK-NEXT: vmov.u8 r0, q1[7]
+; CHECK-NEXT: vmov.16 q3[7], r0
+; CHECK-NEXT: vmov.u8 r0, q0[0]
+; CHECK-NEXT: vmovlb.s8 q1, q3
+; CHECK-NEXT: vmov.16 q3[0], r0
+; CHECK-NEXT: vmov.u8 r0, q0[1]
+; CHECK-NEXT: vmov.16 q3[1], r0
+; CHECK-NEXT: vmov.u8 r0, q0[2]
+; CHECK-NEXT: vmov.16 q3[2], r0
+; CHECK-NEXT: vmov.u8 r0, q0[3]
+; CHECK-NEXT: vmov.16 q3[3], r0
+; CHECK-NEXT: vmov.u8 r0, q0[4]
+; CHECK-NEXT: vmov.16 q3[4], r0
+; CHECK-NEXT: vmov.u8 r0, q0[5]
+; CHECK-NEXT: vmov.16 q3[5], r0
+; CHECK-NEXT: vmov.u8 r0, q0[6]
+; CHECK-NEXT: vmov.16 q3[6], r0
+; CHECK-NEXT: vmov.u8 r0, q0[7]
+; CHECK-NEXT: vmov.16 q3[7], r0
+; CHECK-NEXT: vmovlb.s8 q0, q3
+; CHECK-NEXT: vmul.i16 q0, q0, q1
+; CHECK-NEXT: vadd.i16 q0, q0, q2
+; CHECK-NEXT: vaddv.u16 r0, q0
+; CHECK-NEXT: sxth r0, r0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <16 x i8> %x to <16 x i16>
+ %yy = sext <16 x i8> %y to <16 x i16>
+ %m = mul <16 x i16> %xx, %yy
+ %z = call i16 @llvm.experimental.vector.reduce.add.v16i16(<16 x i16> %m)
+ ret i16 %z
+}
+
+define arm_aapcs_vfpcc zeroext i16 @add_v8i8_v8i16_zext(<8 x i8> %x, <8 x i8> %y) {
+; CHECK-LABEL: add_v8i8_v8i16_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmovlb.u8 q1, q1
+; CHECK-NEXT: vmovlb.u8 q0, q0
+; CHECK-NEXT: vmul.i16 q0, q0, q1
+; CHECK-NEXT: vaddv.u16 r0, q0
+; CHECK-NEXT: uxth r0, r0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = zext <8 x i8> %x to <8 x i16>
+ %yy = zext <8 x i8> %y to <8 x i16>
+ %m = mul <8 x i16> %xx, %yy
+ %z = call i16 @llvm.experimental.vector.reduce.add.v8i16(<8 x i16> %m)
+ ret i16 %z
+}
+
+define arm_aapcs_vfpcc signext i16 @add_v8i8_v8i16_sext(<8 x i8> %x, <8 x i8> %y) {
+; CHECK-LABEL: add_v8i8_v8i16_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmovlb.s8 q1, q1
+; CHECK-NEXT: vmovlb.s8 q0, q0
+; CHECK-NEXT: vmul.i16 q0, q0, q1
+; CHECK-NEXT: vaddv.u16 r0, q0
+; CHECK-NEXT: sxth r0, r0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <8 x i8> %x to <8 x i16>
+ %yy = sext <8 x i8> %y to <8 x i16>
+ %m = mul <8 x i16> %xx, %yy
+ %z = call i16 @llvm.experimental.vector.reduce.add.v8i16(<8 x i16> %m)
+ ret i16 %z
+}
+
+define arm_aapcs_vfpcc zeroext i8 @add_v16i8_v16i8(<16 x i8> %x, <16 x i8> %y) {
+; CHECK-LABEL: add_v16i8_v16i8:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmul.i8 q0, q0, q1
+; CHECK-NEXT: vaddv.u8 r0, q0
+; CHECK-NEXT: uxtb r0, r0
+; CHECK-NEXT: bx lr
+entry:
+ %m = mul <16 x i8> %x, %y
+ %z = call i8 @llvm.experimental.vector.reduce.add.v16i8(<16 x i8> %m)
+ ret i8 %z
+}
+
+define arm_aapcs_vfpcc i64 @add_v16i8_v16i64_zext(<16 x i8> %x, <16 x i8> %y) {
+; CHECK-LABEL: add_v16i8_v16i64_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r7, lr}
+; CHECK-NEXT: push {r7, lr}
+; CHECK-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-NEXT: vmov.u8 r0, q1[0]
+; CHECK-NEXT: vmov.u8 r1, q0[0]
+; CHECK-NEXT: vmov.32 q3[0], r0
+; CHECK-NEXT: vmov.u8 r0, q1[1]
+; CHECK-NEXT: vmov.32 q3[2], r0
+; CHECK-NEXT: adr r0, .LCPI23_0
+; CHECK-NEXT: vldrw.u32 q2, [r0]
+; CHECK-NEXT: vmov.32 q4[0], r1
+; CHECK-NEXT: vmov.u8 r1, q0[1]
+; CHECK-NEXT: vmov.32 q4[2], r1
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r0, s14
+; CHECK-NEXT: vmov r1, s18
+; CHECK-NEXT: vmov r2, s12
+; CHECK-NEXT: vmov r3, s16
+; CHECK-NEXT: umull r12, r1, r1, r0
+; CHECK-NEXT: vmov.u8 r0, q0[2]
+; CHECK-NEXT: vmov.32 q4[0], r0
+; CHECK-NEXT: vmov.u8 r0, q0[3]
+; CHECK-NEXT: vmov.32 q4[2], r0
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r0, s16
+; CHECK-NEXT: orr.w lr, r3, r1
+; CHECK-NEXT: vmov.u8 r3, q1[2]
+; CHECK-NEXT: vmov.32 q3[0], r3
+; CHECK-NEXT: vmov.u8 r3, q1[3]
+; CHECK-NEXT: vmov.32 q3[2], r3
+; CHECK-NEXT: add r2, r12
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vmov r3, s12
+; CHECK-NEXT: umull r0, r3, r0, r3
+; CHECK-NEXT: vmov.32 q5[0], r0
+; CHECK-NEXT: vmov r0, s14
+; CHECK-NEXT: vmov.32 q5[1], r3
+; CHECK-NEXT: vmov r3, s18
+; CHECK-NEXT: umull r0, r3, r3, r0
+; CHECK-NEXT: vmov.32 q5[2], r0
+; CHECK-NEXT: vmov.32 q5[3], r3
+; CHECK-NEXT: vmov r1, s20
+; CHECK-NEXT: vmov r0, s21
+; CHECK-NEXT: adds r1, r1, r2
+; CHECK-NEXT: adc.w r2, lr, r0
+; CHECK-NEXT: vmov r0, s22
+; CHECK-NEXT: adds.w r12, r1, r0
+; CHECK-NEXT: adc.w r1, r2, r3
+; CHECK-NEXT: vmov.u8 r2, q1[4]
+; CHECK-NEXT: vmov.u8 r3, q0[4]
+; CHECK-NEXT: vmov.32 q3[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[5]
+; CHECK-NEXT: vmov.32 q4[0], r3
+; CHECK-NEXT: vmov.u8 r3, q0[5]
+; CHECK-NEXT: vmov.32 q3[2], r2
+; CHECK-NEXT: vmov.32 q4[2], r3
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r2, s12
+; CHECK-NEXT: vmov r3, s16
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[0], r2
+; CHECK-NEXT: vmov r2, s14
+; CHECK-NEXT: vmov.32 q5[1], r3
+; CHECK-NEXT: vmov r3, s18
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[2], r2
+; CHECK-NEXT: vmov.32 q5[3], r3
+; CHECK-NEXT: vmov r0, s20
+; CHECK-NEXT: vmov r2, s21
+; CHECK-NEXT: adds.w r0, r0, r12
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov r2, s22
+; CHECK-NEXT: adds.w r12, r0, r2
+; CHECK-NEXT: vmov.u8 r2, q1[6]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.u8 r3, q0[6]
+; CHECK-NEXT: vmov.32 q3[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[7]
+; CHECK-NEXT: vmov.32 q4[0], r3
+; CHECK-NEXT: vmov.u8 r3, q0[7]
+; CHECK-NEXT: vmov.32 q3[2], r2
+; CHECK-NEXT: vmov.32 q4[2], r3
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r2, s12
+; CHECK-NEXT: vmov r3, s16
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[0], r2
+; CHECK-NEXT: vmov r2, s14
+; CHECK-NEXT: vmov.32 q5[1], r3
+; CHECK-NEXT: vmov r3, s18
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[2], r2
+; CHECK-NEXT: vmov.32 q5[3], r3
+; CHECK-NEXT: vmov r0, s20
+; CHECK-NEXT: vmov r2, s21
+; CHECK-NEXT: adds.w r0, r0, r12
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov r2, s22
+; CHECK-NEXT: adds.w r12, r0, r2
+; CHECK-NEXT: vmov.u8 r2, q1[8]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.u8 r3, q0[8]
+; CHECK-NEXT: vmov.32 q3[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[9]
+; CHECK-NEXT: vmov.32 q4[0], r3
+; CHECK-NEXT: vmov.u8 r3, q0[9]
+; CHECK-NEXT: vmov.32 q3[2], r2
+; CHECK-NEXT: vmov.32 q4[2], r3
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r2, s12
+; CHECK-NEXT: vmov r3, s16
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[0], r2
+; CHECK-NEXT: vmov r2, s14
+; CHECK-NEXT: vmov.32 q5[1], r3
+; CHECK-NEXT: vmov r3, s18
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[2], r2
+; CHECK-NEXT: vmov.32 q5[3], r3
+; CHECK-NEXT: vmov r0, s20
+; CHECK-NEXT: vmov r2, s21
+; CHECK-NEXT: adds.w r0, r0, r12
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov r2, s22
+; CHECK-NEXT: adds.w r12, r0, r2
+; CHECK-NEXT: vmov.u8 r2, q1[10]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.u8 r3, q0[10]
+; CHECK-NEXT: vmov.32 q3[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[11]
+; CHECK-NEXT: vmov.32 q4[0], r3
+; CHECK-NEXT: vmov.u8 r3, q0[11]
+; CHECK-NEXT: vmov.32 q3[2], r2
+; CHECK-NEXT: vmov.32 q4[2], r3
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r2, s12
+; CHECK-NEXT: vmov r3, s16
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[0], r2
+; CHECK-NEXT: vmov r2, s14
+; CHECK-NEXT: vmov.32 q5[1], r3
+; CHECK-NEXT: vmov r3, s18
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[2], r2
+; CHECK-NEXT: vmov.32 q5[3], r3
+; CHECK-NEXT: vmov r0, s20
+; CHECK-NEXT: vmov r2, s21
+; CHECK-NEXT: adds.w r0, r0, r12
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov r2, s22
+; CHECK-NEXT: adds.w r12, r0, r2
+; CHECK-NEXT: vmov.u8 r2, q1[12]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.u8 r3, q0[12]
+; CHECK-NEXT: vmov.32 q3[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[13]
+; CHECK-NEXT: vmov.32 q4[0], r3
+; CHECK-NEXT: vmov.u8 r3, q0[13]
+; CHECK-NEXT: vmov.32 q3[2], r2
+; CHECK-NEXT: vmov.32 q4[2], r3
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r2, s12
+; CHECK-NEXT: vmov r3, s16
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[0], r2
+; CHECK-NEXT: vmov r2, s14
+; CHECK-NEXT: vmov.32 q5[1], r3
+; CHECK-NEXT: vmov r3, s18
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[2], r2
+; CHECK-NEXT: vmov.32 q5[3], r3
+; CHECK-NEXT: vmov r0, s20
+; CHECK-NEXT: vmov r2, s21
+; CHECK-NEXT: adds.w r0, r0, r12
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov r2, s22
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: vmov.u8 r2, q1[14]
+; CHECK-NEXT: vmov.32 q3[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[15]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.32 q3[2], r2
+; CHECK-NEXT: vmov.u8 r3, q0[14]
+; CHECK-NEXT: vand q1, q3, q2
+; CHECK-NEXT: vmov.32 q3[0], r3
+; CHECK-NEXT: vmov.u8 r3, q0[15]
+; CHECK-NEXT: vmov.32 q3[2], r3
+; CHECK-NEXT: vmov r2, s4
+; CHECK-NEXT: vand q0, q3, q2
+; CHECK-NEXT: vmov r3, s0
+; CHECK-NEXT: umlal r0, r1, r3, r2
+; CHECK-NEXT: vmov r2, s6
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: umlal r0, r1, r3, r2
+; CHECK-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-NEXT: pop {r7, pc}
+; CHECK-NEXT: .p2align 4
+; CHECK-NEXT: @ %bb.1:
+; CHECK-NEXT: .LCPI23_0:
+; CHECK-NEXT: .long 255 @ 0xff
+; CHECK-NEXT: .long 0 @ 0x0
+; CHECK-NEXT: .long 255 @ 0xff
+; CHECK-NEXT: .long 0 @ 0x0
+entry:
+ %xx = zext <16 x i8> %x to <16 x i64>
+ %yy = zext <16 x i8> %y to <16 x i64>
+ %m = mul <16 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v16i64(<16 x i64> %m)
+ ret i64 %z
+}
+
+define arm_aapcs_vfpcc i64 @add_v16i8_v16i64_sext(<16 x i8> %x, <16 x i8> %y) {
+; CHECK-LABEL: add_v16i8_v16i64_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov.u8 r0, q1[0]
+; CHECK-NEXT: vmov.u8 r1, q0[0]
+; CHECK-NEXT: sxtb r0, r0
+; CHECK-NEXT: sxtb r1, r1
+; CHECK-NEXT: smull r0, r1, r1, r0
+; CHECK-NEXT: vmov.32 q2[0], r0
+; CHECK-NEXT: vmov.u8 r0, q1[1]
+; CHECK-NEXT: vmov.32 q2[1], r1
+; CHECK-NEXT: vmov.u8 r1, q0[1]
+; CHECK-NEXT: sxtb r0, r0
+; CHECK-NEXT: sxtb r1, r1
+; CHECK-NEXT: smull r0, r1, r1, r0
+; CHECK-NEXT: vmov.32 q2[2], r0
+; CHECK-NEXT: vmov.32 q2[3], r1
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: vmov r3, s8
+; CHECK-NEXT: vmov r0, s9
+; CHECK-NEXT: adds r2, r2, r3
+; CHECK-NEXT: vmov.u8 r3, q0[2]
+; CHECK-NEXT: adc.w r12, r0, r1
+; CHECK-NEXT: vmov.u8 r1, q1[2]
+; CHECK-NEXT: sxtb r1, r1
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r1, r3, r3, r1
+; CHECK-NEXT: vmov.32 q2[0], r1
+; CHECK-NEXT: vmov.u8 r1, q1[3]
+; CHECK-NEXT: vmov.32 q2[1], r3
+; CHECK-NEXT: vmov.u8 r3, q0[3]
+; CHECK-NEXT: sxtb r1, r1
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r1, r3, r3, r1
+; CHECK-NEXT: vmov.32 q2[2], r1
+; CHECK-NEXT: vmov.32 q2[3], r3
+; CHECK-NEXT: vmov r0, s8
+; CHECK-NEXT: vmov r1, s9
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adc.w r1, r1, r12
+; CHECK-NEXT: adds.w r12, r0, r2
+; CHECK-NEXT: vmov.u8 r2, q1[4]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.u8 r3, q0[4]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[5]
+; CHECK-NEXT: vmov.32 q2[1], r3
+; CHECK-NEXT: vmov.u8 r3, q0[5]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r3
+; CHECK-NEXT: vmov r0, s8
+; CHECK-NEXT: vmov r2, s9
+; CHECK-NEXT: adds.w r0, r0, r12
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adds.w r12, r0, r2
+; CHECK-NEXT: vmov.u8 r2, q1[6]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.u8 r3, q0[6]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[7]
+; CHECK-NEXT: vmov.32 q2[1], r3
+; CHECK-NEXT: vmov.u8 r3, q0[7]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r3
+; CHECK-NEXT: vmov r0, s8
+; CHECK-NEXT: vmov r2, s9
+; CHECK-NEXT: adds.w r0, r0, r12
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adds.w r12, r0, r2
+; CHECK-NEXT: vmov.u8 r2, q1[8]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.u8 r3, q0[8]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[9]
+; CHECK-NEXT: vmov.32 q2[1], r3
+; CHECK-NEXT: vmov.u8 r3, q0[9]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r3
+; CHECK-NEXT: vmov r0, s8
+; CHECK-NEXT: vmov r2, s9
+; CHECK-NEXT: adds.w r0, r0, r12
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adds.w r12, r0, r2
+; CHECK-NEXT: vmov.u8 r2, q1[10]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.u8 r3, q0[10]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[11]
+; CHECK-NEXT: vmov.32 q2[1], r3
+; CHECK-NEXT: vmov.u8 r3, q0[11]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r3
+; CHECK-NEXT: vmov r0, s8
+; CHECK-NEXT: vmov r2, s9
+; CHECK-NEXT: adds.w r0, r0, r12
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adds.w r12, r0, r2
+; CHECK-NEXT: vmov.u8 r2, q1[12]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.u8 r3, q0[12]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[13]
+; CHECK-NEXT: vmov.32 q2[1], r3
+; CHECK-NEXT: vmov.u8 r3, q0[13]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r3
+; CHECK-NEXT: vmov r0, s8
+; CHECK-NEXT: vmov r2, s9
+; CHECK-NEXT: adds.w r0, r0, r12
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: vmov.u8 r2, q1[14]
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vmov.u8 r3, q0[14]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smlal r0, r1, r3, r2
+; CHECK-NEXT: vmov.u8 r2, q1[15]
+; CHECK-NEXT: vmov.u8 r3, q0[15]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smlal r0, r1, r3, r2
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <16 x i8> %x to <16 x i64>
+ %yy = sext <16 x i8> %y to <16 x i64>
+ %m = mul <16 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v16i64(<16 x i64> %m)
+ ret i64 %z
+}
+
+define arm_aapcs_vfpcc i64 @add_v2i8_v2i64_zext(<2 x i8> %x, <2 x i8> %y) {
+; CHECK-LABEL: add_v2i8_v2i64_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: adr r0, .LCPI25_0
+; CHECK-NEXT: vldrw.u32 q2, [r0]
+; CHECK-NEXT: vand q1, q1, q2
+; CHECK-NEXT: vand q0, q0, q2
+; CHECK-NEXT: vmov r0, s6
+; CHECK-NEXT: vmov r1, s2
+; CHECK-NEXT: vmov r2, s4
+; CHECK-NEXT: vmov r3, s0
+; CHECK-NEXT: umull r0, r1, r1, r0
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: add r0, r2
+; CHECK-NEXT: orrs r1, r3
+; CHECK-NEXT: bx lr
+; CHECK-NEXT: .p2align 4
+; CHECK-NEXT: @ %bb.1:
+; CHECK-NEXT: .LCPI25_0:
+; CHECK-NEXT: .long 255 @ 0xff
+; CHECK-NEXT: .long 0 @ 0x0
+; CHECK-NEXT: .long 255 @ 0xff
+; CHECK-NEXT: .long 0 @ 0x0
+entry:
+ %xx = zext <2 x i8> %x to <2 x i64>
+ %yy = zext <2 x i8> %y to <2 x i64>
+ %m = mul <2 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64> %m)
+ ret i64 %z
+}
+
+define arm_aapcs_vfpcc i64 @add_v2i8_v2i64_sext(<2 x i8> %x, <2 x i8> %y) {
+; CHECK-LABEL: add_v2i8_v2i64_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov r0, s4
+; CHECK-NEXT: vmov r1, s0
+; CHECK-NEXT: vmov r2, s6
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: sxtb r0, r0
+; CHECK-NEXT: sxtb r1, r1
+; CHECK-NEXT: smull r0, r1, r1, r0
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smlal r0, r1, r3, r2
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <2 x i8> %x to <2 x i64>
+ %yy = sext <2 x i8> %y to <2 x i64>
+ %m = mul <2 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64> %m)
+ ret i64 %z
+}
+
+define arm_aapcs_vfpcc i64 @add_v2i64_v2i64(<2 x i64> %x, <2 x i64> %y) {
+; CHECK-LABEL: add_v2i64_v2i64:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r4, lr}
+; CHECK-NEXT: push {r4, lr}
+; CHECK-NEXT: vmov r0, s4
+; CHECK-NEXT: vmov r1, s0
+; CHECK-NEXT: vmov r2, s5
+; CHECK-NEXT: vmov r4, s7
+; CHECK-NEXT: umull r12, r3, r1, r0
+; CHECK-NEXT: mla r1, r1, r2, r3
+; CHECK-NEXT: vmov r2, s1
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: vmov.32 q2[0], r12
+; CHECK-NEXT: mla r1, r2, r0, r1
+; CHECK-NEXT: vmov r2, s6
+; CHECK-NEXT: vmov.32 q2[1], r1
+; CHECK-NEXT: vmov r12, s8
+; CHECK-NEXT: umull lr, r0, r3, r2
+; CHECK-NEXT: mla r0, r3, r4, r0
+; CHECK-NEXT: vmov r3, s3
+; CHECK-NEXT: mla r2, r3, r2, r0
+; CHECK-NEXT: adds.w r0, r12, lr
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: pop {r4, pc}
+entry:
+ %m = mul <2 x i64> %x, %y
+ %z = call i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64> %m)
+ ret i64 %z
+}
+
+define arm_aapcs_vfpcc i32 @add_v4i32_v4i32_acc(<4 x i32> %x, <4 x i32> %y, i32 %a) {
+; CHECK-LABEL: add_v4i32_v4i32_acc:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vaddva.u32 r0, q0
+; CHECK-NEXT: bx lr
+entry:
+ %m = mul <4 x i32> %x, %y
+ %z = call i32 @llvm.experimental.vector.reduce.add.v4i32(<4 x i32> %m)
+ %r = add i32 %z, %a
+ ret i32 %r
+}
+
+define arm_aapcs_vfpcc i64 @add_v4i32_v4i64_acc_zext(<4 x i32> %x, <4 x i32> %y, i64 %a) {
+; CHECK-LABEL: add_v4i32_v4i64_acc_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r7, lr}
+; CHECK-NEXT: push {r7, lr}
+; CHECK-NEXT: .vsave {d8, d9}
+; CHECK-NEXT: vpush {d8, d9}
+; CHECK-NEXT: vmov.f32 s8, s4
+; CHECK-NEXT: vmov.f32 s12, s0
+; CHECK-NEXT: vmov.f32 s10, s5
+; CHECK-NEXT: vmov.f32 s14, s1
+; CHECK-NEXT: vmov r2, s8
+; CHECK-NEXT: vmov r3, s12
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q4[0], r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: vmov.32 q4[1], r3
+; CHECK-NEXT: vmov r3, s14
+; CHECK-NEXT: vmov.f32 s8, s6
+; CHECK-NEXT: vmov.f32 s12, s2
+; CHECK-NEXT: vmov.f32 s10, s7
+; CHECK-NEXT: vmov.f32 s14, s3
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q4[2], r2
+; CHECK-NEXT: vmov.32 q4[3], r3
+; CHECK-NEXT: vmov lr, s18
+; CHECK-NEXT: vmov r2, s16
+; CHECK-NEXT: vmov r12, s17
+; CHECK-NEXT: adds.w lr, lr, r2
+; CHECK-NEXT: adr r2, .LCPI29_0
+; CHECK-NEXT: vldrw.u32 q1, [r2]
+; CHECK-NEXT: adc.w r3, r3, r12
+; CHECK-NEXT: vand q2, q2, q1
+; CHECK-NEXT: vand q0, q3, q1
+; CHECK-NEXT: vmov r12, s8
+; CHECK-NEXT: vmov r2, s0
+; CHECK-NEXT: umlal lr, r3, r2, r12
+; CHECK-NEXT: vmov r12, s10
+; CHECK-NEXT: vmov r2, s2
+; CHECK-NEXT: umlal lr, r3, r2, r12
+; CHECK-NEXT: adds.w r0, r0, lr
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vpop {d8, d9}
+; CHECK-NEXT: pop {r7, pc}
+; CHECK-NEXT: .p2align 4
+; CHECK-NEXT: @ %bb.1:
+; CHECK-NEXT: .LCPI29_0:
+; CHECK-NEXT: .long 4294967295 @ 0xffffffff
+; CHECK-NEXT: .long 0 @ 0x0
+; CHECK-NEXT: .long 4294967295 @ 0xffffffff
+; CHECK-NEXT: .long 0 @ 0x0
+entry:
+ %xx = zext <4 x i32> %x to <4 x i64>
+ %yy = zext <4 x i32> %y to <4 x i64>
+ %m = mul <4 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v4i64(<4 x i64> %m)
+ %r = add i64 %z, %a
+ ret i64 %r
+}
+
+define arm_aapcs_vfpcc i64 @add_v4i32_v4i64_acc_sext(<4 x i32> %x, <4 x i32> %y, i64 %a) {
+; CHECK-LABEL: add_v4i32_v4i64_acc_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r7, lr}
+; CHECK-NEXT: push {r7, lr}
+; CHECK-NEXT: .vsave {d8, d9}
+; CHECK-NEXT: vpush {d8, d9}
+; CHECK-NEXT: vmov.f32 s8, s4
+; CHECK-NEXT: vmov.f32 s12, s0
+; CHECK-NEXT: vmov.f32 s10, s5
+; CHECK-NEXT: vmov.f32 s14, s1
+; CHECK-NEXT: vmov r2, s8
+; CHECK-NEXT: vmov r3, s12
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q4[0], r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: vmov.32 q4[1], r3
+; CHECK-NEXT: vmov r3, s14
+; CHECK-NEXT: vmov.f32 s8, s6
+; CHECK-NEXT: vmov.f32 s10, s7
+; CHECK-NEXT: vmov.f32 s4, s2
+; CHECK-NEXT: vmov.f32 s6, s3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q4[2], r2
+; CHECK-NEXT: vmov.32 q4[3], r3
+; CHECK-NEXT: vmov lr, s18
+; CHECK-NEXT: vmov r2, s16
+; CHECK-NEXT: vmov r12, s17
+; CHECK-NEXT: adds.w lr, lr, r2
+; CHECK-NEXT: vmov r2, s4
+; CHECK-NEXT: adc.w r3, r3, r12
+; CHECK-NEXT: vmov r12, s8
+; CHECK-NEXT: smlal lr, r3, r2, r12
+; CHECK-NEXT: vmov r12, s10
+; CHECK-NEXT: vmov r2, s6
+; CHECK-NEXT: smlal lr, r3, r2, r12
+; CHECK-NEXT: adds.w r0, r0, lr
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vpop {d8, d9}
+; CHECK-NEXT: pop {r7, pc}
+entry:
+ %xx = sext <4 x i32> %x to <4 x i64>
+ %yy = sext <4 x i32> %y to <4 x i64>
+ %m = mul <4 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v4i64(<4 x i64> %m)
+ %r = add i64 %z, %a
+ ret i64 %r
+}
+
+define arm_aapcs_vfpcc i64 @add_v2i32_v2i64_acc_zext(<2 x i32> %x, <2 x i32> %y, i64 %a) {
+; CHECK-LABEL: add_v2i32_v2i64_acc_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r7, lr}
+; CHECK-NEXT: push {r7, lr}
+; CHECK-NEXT: vmov r2, s4
+; CHECK-NEXT: vmov r3, s0
+; CHECK-NEXT: vmov r12, s6
+; CHECK-NEXT: umull r2, lr, r3, r2
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: umlal r2, lr, r3, r12
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: adc.w r1, r1, lr
+; CHECK-NEXT: pop {r7, pc}
+entry:
+ %xx = zext <2 x i32> %x to <2 x i64>
+ %yy = zext <2 x i32> %y to <2 x i64>
+ %m = mul <2 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64> %m)
+ %r = add i64 %z, %a
+ ret i64 %r
+}
+
+define arm_aapcs_vfpcc i64 @add_v2i32_v2i64_acc_sext(<2 x i32> %x, <2 x i32> %y, i64 %a) {
+; CHECK-LABEL: add_v2i32_v2i64_acc_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r7, lr}
+; CHECK-NEXT: push {r7, lr}
+; CHECK-NEXT: vmov r2, s4
+; CHECK-NEXT: vmov r3, s0
+; CHECK-NEXT: vmov r12, s6
+; CHECK-NEXT: smull r2, lr, r3, r2
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: smlal r2, lr, r3, r12
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: adc.w r1, r1, lr
+; CHECK-NEXT: pop {r7, pc}
+entry:
+ %xx = sext <2 x i32> %x to <2 x i64>
+ %yy = sext <2 x i32> %y to <2 x i64>
+ %m = mul <2 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64> %m)
+ %r = add i64 %z, %a
+ ret i64 %r
+}
+
+define arm_aapcs_vfpcc i32 @add_v8i16_v8i32_acc_zext(<8 x i16> %x, <8 x i16> %y, i32 %a) {
+; CHECK-LABEL: add_v8i16_v8i32_acc_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov.u16 r1, q1[4]
+; CHECK-NEXT: vmov.32 q2[0], r1
+; CHECK-NEXT: vmov.u16 r1, q1[5]
+; CHECK-NEXT: vmov.32 q2[1], r1
+; CHECK-NEXT: vmov.u16 r1, q1[6]
+; CHECK-NEXT: vmov.32 q2[2], r1
+; CHECK-NEXT: vmov.u16 r1, q1[7]
+; CHECK-NEXT: vmov.32 q2[3], r1
+; CHECK-NEXT: vmov.u16 r1, q0[4]
+; CHECK-NEXT: vmov.32 q3[0], r1
+; CHECK-NEXT: vmov.u16 r1, q0[5]
+; CHECK-NEXT: vmov.32 q3[1], r1
+; CHECK-NEXT: vmov.u16 r1, q0[6]
+; CHECK-NEXT: vmov.32 q3[2], r1
+; CHECK-NEXT: vmov.u16 r1, q0[7]
+; CHECK-NEXT: vmov.32 q3[3], r1
+; CHECK-NEXT: vmovlb.u16 q2, q2
+; CHECK-NEXT: vmovlb.u16 q3, q3
+; CHECK-NEXT: vmov.u16 r1, q1[0]
+; CHECK-NEXT: vmul.i32 q2, q3, q2
+; CHECK-NEXT: vmov.32 q3[0], r1
+; CHECK-NEXT: vmov.u16 r1, q1[1]
+; CHECK-NEXT: vmov.32 q3[1], r1
+; CHECK-NEXT: vmov.u16 r1, q1[2]
+; CHECK-NEXT: vmov.32 q3[2], r1
+; CHECK-NEXT: vmov.u16 r1, q1[3]
+; CHECK-NEXT: vmov.32 q3[3], r1
+; CHECK-NEXT: vmov.u16 r1, q0[0]
+; CHECK-NEXT: vmovlb.u16 q1, q3
+; CHECK-NEXT: vmov.32 q3[0], r1
+; CHECK-NEXT: vmov.u16 r1, q0[1]
+; CHECK-NEXT: vmov.32 q3[1], r1
+; CHECK-NEXT: vmov.u16 r1, q0[2]
+; CHECK-NEXT: vmov.32 q3[2], r1
+; CHECK-NEXT: vmov.u16 r1, q0[3]
+; CHECK-NEXT: vmov.32 q3[3], r1
+; CHECK-NEXT: vmovlb.u16 q0, q3
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vadd.i32 q0, q0, q2
+; CHECK-NEXT: vaddva.u32 r0, q0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = zext <8 x i16> %x to <8 x i32>
+ %yy = zext <8 x i16> %y to <8 x i32>
+ %m = mul <8 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v8i32(<8 x i32> %m)
+ %r = add i32 %z, %a
+ ret i32 %r
+}
+
+define arm_aapcs_vfpcc i32 @add_v8i16_v8i32_acc_sext(<8 x i16> %x, <8 x i16> %y, i32 %a) {
+; CHECK-LABEL: add_v8i16_v8i32_acc_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov.u16 r1, q1[4]
+; CHECK-NEXT: vmov.32 q2[0], r1
+; CHECK-NEXT: vmov.u16 r1, q1[5]
+; CHECK-NEXT: vmov.32 q2[1], r1
+; CHECK-NEXT: vmov.u16 r1, q1[6]
+; CHECK-NEXT: vmov.32 q2[2], r1
+; CHECK-NEXT: vmov.u16 r1, q1[7]
+; CHECK-NEXT: vmov.32 q2[3], r1
+; CHECK-NEXT: vmov.u16 r1, q0[4]
+; CHECK-NEXT: vmov.32 q3[0], r1
+; CHECK-NEXT: vmov.u16 r1, q0[5]
+; CHECK-NEXT: vmov.32 q3[1], r1
+; CHECK-NEXT: vmov.u16 r1, q0[6]
+; CHECK-NEXT: vmov.32 q3[2], r1
+; CHECK-NEXT: vmov.u16 r1, q0[7]
+; CHECK-NEXT: vmov.32 q3[3], r1
+; CHECK-NEXT: vmovlb.s16 q2, q2
+; CHECK-NEXT: vmovlb.s16 q3, q3
+; CHECK-NEXT: vmov.u16 r1, q1[0]
+; CHECK-NEXT: vmul.i32 q2, q3, q2
+; CHECK-NEXT: vmov.32 q3[0], r1
+; CHECK-NEXT: vmov.u16 r1, q1[1]
+; CHECK-NEXT: vmov.32 q3[1], r1
+; CHECK-NEXT: vmov.u16 r1, q1[2]
+; CHECK-NEXT: vmov.32 q3[2], r1
+; CHECK-NEXT: vmov.u16 r1, q1[3]
+; CHECK-NEXT: vmov.32 q3[3], r1
+; CHECK-NEXT: vmov.u16 r1, q0[0]
+; CHECK-NEXT: vmovlb.s16 q1, q3
+; CHECK-NEXT: vmov.32 q3[0], r1
+; CHECK-NEXT: vmov.u16 r1, q0[1]
+; CHECK-NEXT: vmov.32 q3[1], r1
+; CHECK-NEXT: vmov.u16 r1, q0[2]
+; CHECK-NEXT: vmov.32 q3[2], r1
+; CHECK-NEXT: vmov.u16 r1, q0[3]
+; CHECK-NEXT: vmov.32 q3[3], r1
+; CHECK-NEXT: vmovlb.s16 q0, q3
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vadd.i32 q0, q0, q2
+; CHECK-NEXT: vaddva.u32 r0, q0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <8 x i16> %x to <8 x i32>
+ %yy = sext <8 x i16> %y to <8 x i32>
+ %m = mul <8 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v8i32(<8 x i32> %m)
+ %r = add i32 %z, %a
+ ret i32 %r
+}
+
+define arm_aapcs_vfpcc i32 @add_v4i16_v4i32_acc_zext(<4 x i16> %x, <4 x i16> %y, i32 %a) {
+; CHECK-LABEL: add_v4i16_v4i32_acc_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmovlb.u16 q1, q1
+; CHECK-NEXT: vmovlb.u16 q0, q0
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vaddva.u32 r0, q0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = zext <4 x i16> %x to <4 x i32>
+ %yy = zext <4 x i16> %y to <4 x i32>
+ %m = mul <4 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v4i32(<4 x i32> %m)
+ %r = add i32 %z, %a
+ ret i32 %r
+}
+
+define arm_aapcs_vfpcc i32 @add_v4i16_v4i32_acc_sext(<4 x i16> %x, <4 x i16> %y, i32 %a) {
+; CHECK-LABEL: add_v4i16_v4i32_acc_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmovlb.s16 q1, q1
+; CHECK-NEXT: vmovlb.s16 q0, q0
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vaddva.u32 r0, q0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <4 x i16> %x to <4 x i32>
+ %yy = sext <4 x i16> %y to <4 x i32>
+ %m = mul <4 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v4i32(<4 x i32> %m)
+ %r = add i32 %z, %a
+ ret i32 %r
+}
+
+define arm_aapcs_vfpcc zeroext i16 @add_v8i16_v8i16_acc(<8 x i16> %x, <8 x i16> %y, i16 %a) {
+; CHECK-LABEL: add_v8i16_v8i16_acc:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmul.i16 q0, q0, q1
+; CHECK-NEXT: vaddva.u16 r0, q0
+; CHECK-NEXT: uxth r0, r0
+; CHECK-NEXT: bx lr
+entry:
+ %m = mul <8 x i16> %x, %y
+ %z = call i16 @llvm.experimental.vector.reduce.add.v8i16(<8 x i16> %m)
+ %r = add i16 %z, %a
+ ret i16 %r
+}
+
+define arm_aapcs_vfpcc i64 @add_v8i16_v8i64_acc_zext(<8 x i16> %x, <8 x i16> %y, i64 %a) {
+; CHECK-LABEL: add_v8i16_v8i64_acc_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r4, lr}
+; CHECK-NEXT: push {r4, lr}
+; CHECK-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-NEXT: vmov.u16 r2, q1[0]
+; CHECK-NEXT: vmov.u16 r3, q0[0]
+; CHECK-NEXT: vmov.32 q3[0], r2
+; CHECK-NEXT: vmov.u16 r2, q1[1]
+; CHECK-NEXT: vmov.32 q3[2], r2
+; CHECK-NEXT: adr r2, .LCPI38_0
+; CHECK-NEXT: vldrw.u32 q2, [r2]
+; CHECK-NEXT: vmov.32 q4[0], r3
+; CHECK-NEXT: vmov.u16 r3, q0[1]
+; CHECK-NEXT: vmov.32 q4[2], r3
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r2, s12
+; CHECK-NEXT: vmov r3, s16
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[0], r2
+; CHECK-NEXT: vmov r2, s14
+; CHECK-NEXT: vmov.32 q5[1], r3
+; CHECK-NEXT: vmov r3, s18
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[2], r2
+; CHECK-NEXT: vmov.32 q5[3], r3
+; CHECK-NEXT: vmov r12, s22
+; CHECK-NEXT: vmov r2, s20
+; CHECK-NEXT: vmov lr, s21
+; CHECK-NEXT: adds.w r12, r12, r2
+; CHECK-NEXT: vmov.u16 r2, q1[2]
+; CHECK-NEXT: adc.w lr, lr, r3
+; CHECK-NEXT: vmov.u16 r3, q0[2]
+; CHECK-NEXT: vmov.32 q3[0], r2
+; CHECK-NEXT: vmov.u16 r2, q1[3]
+; CHECK-NEXT: vmov.32 q4[0], r3
+; CHECK-NEXT: vmov.u16 r3, q0[3]
+; CHECK-NEXT: vmov.32 q3[2], r2
+; CHECK-NEXT: vmov.32 q4[2], r3
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r2, s12
+; CHECK-NEXT: vmov r3, s16
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[0], r2
+; CHECK-NEXT: vmov r2, s14
+; CHECK-NEXT: vmov.32 q5[1], r3
+; CHECK-NEXT: vmov r3, s18
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[2], r2
+; CHECK-NEXT: vmov.32 q5[3], r3
+; CHECK-NEXT: vmov r4, s20
+; CHECK-NEXT: vmov r2, s21
+; CHECK-NEXT: adds.w r4, r4, r12
+; CHECK-NEXT: adc.w r12, lr, r2
+; CHECK-NEXT: vmov r2, s22
+; CHECK-NEXT: adds.w lr, r4, r2
+; CHECK-NEXT: vmov.u16 r2, q1[4]
+; CHECK-NEXT: adc.w r12, r12, r3
+; CHECK-NEXT: vmov.u16 r3, q0[4]
+; CHECK-NEXT: vmov.32 q3[0], r2
+; CHECK-NEXT: vmov.u16 r2, q1[5]
+; CHECK-NEXT: vmov.32 q4[0], r3
+; CHECK-NEXT: vmov.u16 r3, q0[5]
+; CHECK-NEXT: vmov.32 q3[2], r2
+; CHECK-NEXT: vmov.32 q4[2], r3
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r2, s12
+; CHECK-NEXT: vmov r3, s16
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[0], r2
+; CHECK-NEXT: vmov r2, s14
+; CHECK-NEXT: vmov.32 q5[1], r3
+; CHECK-NEXT: vmov r3, s18
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q5[2], r2
+; CHECK-NEXT: vmov.32 q5[3], r3
+; CHECK-NEXT: vmov r4, s20
+; CHECK-NEXT: vmov r2, s21
+; CHECK-NEXT: adds.w r4, r4, lr
+; CHECK-NEXT: adc.w r12, r12, r2
+; CHECK-NEXT: vmov r2, s22
+; CHECK-NEXT: adds r2, r2, r4
+; CHECK-NEXT: vmov.u16 r4, q1[6]
+; CHECK-NEXT: vmov.32 q3[0], r4
+; CHECK-NEXT: vmov.u16 r4, q1[7]
+; CHECK-NEXT: vmov.32 q3[2], r4
+; CHECK-NEXT: vmov.u16 r4, q0[6]
+; CHECK-NEXT: vand q1, q3, q2
+; CHECK-NEXT: vmov.32 q3[0], r4
+; CHECK-NEXT: vmov.u16 r4, q0[7]
+; CHECK-NEXT: adc.w r3, r3, r12
+; CHECK-NEXT: vmov.32 q3[2], r4
+; CHECK-NEXT: vmov r12, s4
+; CHECK-NEXT: vand q0, q3, q2
+; CHECK-NEXT: vmov r4, s0
+; CHECK-NEXT: umlal r2, r3, r4, r12
+; CHECK-NEXT: vmov r12, s6
+; CHECK-NEXT: vmov r4, s2
+; CHECK-NEXT: umlal r2, r3, r4, r12
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-NEXT: pop {r4, pc}
+; CHECK-NEXT: .p2align 4
+; CHECK-NEXT: @ %bb.1:
+; CHECK-NEXT: .LCPI38_0:
+; CHECK-NEXT: .long 65535 @ 0xffff
+; CHECK-NEXT: .long 0 @ 0x0
+; CHECK-NEXT: .long 65535 @ 0xffff
+; CHECK-NEXT: .long 0 @ 0x0
+entry:
+ %xx = zext <8 x i16> %x to <8 x i64>
+ %yy = zext <8 x i16> %y to <8 x i64>
+ %m = mul <8 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v8i64(<8 x i64> %m)
+ %r = add i64 %z, %a
+ ret i64 %r
+}
+
+define arm_aapcs_vfpcc i64 @add_v8i16_v8i64_acc_sext(<8 x i16> %x, <8 x i16> %y, i64 %a) {
+; CHECK-LABEL: add_v8i16_v8i64_acc_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r4, lr}
+; CHECK-NEXT: push {r4, lr}
+; CHECK-NEXT: vmov.u16 r2, q1[0]
+; CHECK-NEXT: vmov.u16 r3, q0[0]
+; CHECK-NEXT: sxth r2, r2
+; CHECK-NEXT: sxth r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u16 r2, q1[1]
+; CHECK-NEXT: vmov.32 q2[1], r3
+; CHECK-NEXT: vmov.u16 r3, q0[1]
+; CHECK-NEXT: sxth r2, r2
+; CHECK-NEXT: sxth r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r3
+; CHECK-NEXT: vmov lr, s10
+; CHECK-NEXT: vmov r2, s8
+; CHECK-NEXT: vmov r12, s9
+; CHECK-NEXT: adds.w lr, lr, r2
+; CHECK-NEXT: vmov.u16 r2, q1[2]
+; CHECK-NEXT: adc.w r12, r12, r3
+; CHECK-NEXT: vmov.u16 r3, q0[2]
+; CHECK-NEXT: sxth r2, r2
+; CHECK-NEXT: sxth r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u16 r2, q1[3]
+; CHECK-NEXT: vmov.32 q2[1], r3
+; CHECK-NEXT: vmov.u16 r3, q0[3]
+; CHECK-NEXT: sxth r2, r2
+; CHECK-NEXT: sxth r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r3
+; CHECK-NEXT: vmov r4, s8
+; CHECK-NEXT: vmov r2, s9
+; CHECK-NEXT: adds.w r4, r4, lr
+; CHECK-NEXT: adc.w r12, r12, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adds.w lr, r4, r2
+; CHECK-NEXT: vmov.u16 r4, q1[4]
+; CHECK-NEXT: vmov.u16 r2, q0[4]
+; CHECK-NEXT: sxth r4, r4
+; CHECK-NEXT: sxth r2, r2
+; CHECK-NEXT: adc.w r12, r12, r3
+; CHECK-NEXT: smull r2, r4, r2, r4
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u16 r2, q1[5]
+; CHECK-NEXT: vmov.32 q2[1], r4
+; CHECK-NEXT: vmov.u16 r4, q0[5]
+; CHECK-NEXT: sxth r2, r2
+; CHECK-NEXT: sxth r4, r4
+; CHECK-NEXT: smull r2, r4, r4, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r4
+; CHECK-NEXT: vmov r3, s8
+; CHECK-NEXT: vmov r2, s9
+; CHECK-NEXT: adds.w r3, r3, lr
+; CHECK-NEXT: adc.w r12, r12, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adds r2, r2, r3
+; CHECK-NEXT: adc.w r3, r12, r4
+; CHECK-NEXT: vmov.u16 r4, q1[6]
+; CHECK-NEXT: sxth.w r12, r4
+; CHECK-NEXT: vmov.u16 r4, q0[6]
+; CHECK-NEXT: sxth r4, r4
+; CHECK-NEXT: smlal r2, r3, r4, r12
+; CHECK-NEXT: vmov.u16 r4, q1[7]
+; CHECK-NEXT: sxth.w r12, r4
+; CHECK-NEXT: vmov.u16 r4, q0[7]
+; CHECK-NEXT: sxth r4, r4
+; CHECK-NEXT: smlal r2, r3, r4, r12
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: pop {r4, pc}
+entry:
+ %xx = sext <8 x i16> %x to <8 x i64>
+ %yy = sext <8 x i16> %y to <8 x i64>
+ %m = mul <8 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v8i64(<8 x i64> %m)
+ %r = add i64 %z, %a
+ ret i64 %r
+}
+
+define arm_aapcs_vfpcc i64 @add_v2i16_v2i64_acc_zext(<2 x i16> %x, <2 x i16> %y, i64 %a) {
+; CHECK-LABEL: add_v2i16_v2i64_acc_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r7, lr}
+; CHECK-NEXT: push {r7, lr}
+; CHECK-NEXT: adr r2, .LCPI40_0
+; CHECK-NEXT: vldrw.u32 q2, [r2]
+; CHECK-NEXT: vand q1, q1, q2
+; CHECK-NEXT: vand q0, q0, q2
+; CHECK-NEXT: vmov r2, s4
+; CHECK-NEXT: vmov r3, s0
+; CHECK-NEXT: vmov r12, s6
+; CHECK-NEXT: umull r2, lr, r3, r2
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: umlal r2, lr, r3, r12
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: adc.w r1, r1, lr
+; CHECK-NEXT: pop {r7, pc}
+; CHECK-NEXT: .p2align 4
+; CHECK-NEXT: @ %bb.1:
+; CHECK-NEXT: .LCPI40_0:
+; CHECK-NEXT: .long 65535 @ 0xffff
+; CHECK-NEXT: .long 0 @ 0x0
+; CHECK-NEXT: .long 65535 @ 0xffff
+; CHECK-NEXT: .long 0 @ 0x0
+entry:
+ %xx = zext <2 x i16> %x to <2 x i64>
+ %yy = zext <2 x i16> %y to <2 x i64>
+ %m = mul <2 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64> %m)
+ %r = add i64 %z, %a
+ ret i64 %r
+}
+
+define arm_aapcs_vfpcc i64 @add_v2i16_v2i64_acc_sext(<2 x i16> %x, <2 x i16> %y, i64 %a) {
+; CHECK-LABEL: add_v2i16_v2i64_acc_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r7, lr}
+; CHECK-NEXT: push {r7, lr}
+; CHECK-NEXT: vmov r2, s4
+; CHECK-NEXT: vmov r3, s0
+; CHECK-NEXT: sxth r2, r2
+; CHECK-NEXT: sxth r3, r3
+; CHECK-NEXT: smull r2, r12, r3, r2
+; CHECK-NEXT: vmov r3, s6
+; CHECK-NEXT: sxth.w lr, r3
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: sxth r3, r3
+; CHECK-NEXT: smlal r2, r12, r3, lr
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: adc.w r1, r1, r12
+; CHECK-NEXT: pop {r7, pc}
+entry:
+ %xx = sext <2 x i16> %x to <2 x i64>
+ %yy = sext <2 x i16> %y to <2 x i64>
+ %m = mul <2 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64> %m)
+ %r = add i64 %z, %a
+ ret i64 %r
+}
+
+define arm_aapcs_vfpcc i32 @add_v16i8_v16i32_acc_zext(<16 x i8> %x, <16 x i8> %y, i32 %a) {
+; CHECK-LABEL: add_v16i8_v16i32_acc_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-NEXT: vmov.u8 r1, q1[12]
+; CHECK-NEXT: vmov.i32 q2, #0xff
+; CHECK-NEXT: vmov.32 q3[0], r1
+; CHECK-NEXT: vmov.u8 r1, q1[13]
+; CHECK-NEXT: vmov.32 q3[1], r1
+; CHECK-NEXT: vmov.u8 r1, q1[14]
+; CHECK-NEXT: vmov.32 q3[2], r1
+; CHECK-NEXT: vmov.u8 r1, q1[15]
+; CHECK-NEXT: vmov.32 q3[3], r1
+; CHECK-NEXT: vmov.u8 r1, q0[12]
+; CHECK-NEXT: vmov.32 q4[0], r1
+; CHECK-NEXT: vmov.u8 r1, q0[13]
+; CHECK-NEXT: vmov.32 q4[1], r1
+; CHECK-NEXT: vmov.u8 r1, q0[14]
+; CHECK-NEXT: vmov.32 q4[2], r1
+; CHECK-NEXT: vmov.u8 r1, q0[15]
+; CHECK-NEXT: vmov.32 q4[3], r1
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov.u8 r1, q1[4]
+; CHECK-NEXT: vmul.i32 q3, q4, q3
+; CHECK-NEXT: vmov.32 q4[0], r1
+; CHECK-NEXT: vmov.u8 r1, q1[5]
+; CHECK-NEXT: vmov.32 q4[1], r1
+; CHECK-NEXT: vmov.u8 r1, q1[6]
+; CHECK-NEXT: vmov.32 q4[2], r1
+; CHECK-NEXT: vmov.u8 r1, q1[7]
+; CHECK-NEXT: vmov.32 q4[3], r1
+; CHECK-NEXT: vmov.u8 r1, q0[4]
+; CHECK-NEXT: vmov.32 q5[0], r1
+; CHECK-NEXT: vmov.u8 r1, q0[5]
+; CHECK-NEXT: vmov.32 q5[1], r1
+; CHECK-NEXT: vmov.u8 r1, q0[6]
+; CHECK-NEXT: vmov.32 q5[2], r1
+; CHECK-NEXT: vmov.u8 r1, q0[7]
+; CHECK-NEXT: vmov.32 q5[3], r1
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vand q5, q5, q2
+; CHECK-NEXT: vmov.u8 r1, q1[8]
+; CHECK-NEXT: vmul.i32 q4, q5, q4
+; CHECK-NEXT: vadd.i32 q3, q4, q3
+; CHECK-NEXT: vmov.32 q4[0], r1
+; CHECK-NEXT: vmov.u8 r1, q1[9]
+; CHECK-NEXT: vmov.32 q4[1], r1
+; CHECK-NEXT: vmov.u8 r1, q1[10]
+; CHECK-NEXT: vmov.32 q4[2], r1
+; CHECK-NEXT: vmov.u8 r1, q1[11]
+; CHECK-NEXT: vmov.32 q4[3], r1
+; CHECK-NEXT: vmov.u8 r1, q0[8]
+; CHECK-NEXT: vmov.32 q5[0], r1
+; CHECK-NEXT: vmov.u8 r1, q0[9]
+; CHECK-NEXT: vmov.32 q5[1], r1
+; CHECK-NEXT: vmov.u8 r1, q0[10]
+; CHECK-NEXT: vmov.32 q5[2], r1
+; CHECK-NEXT: vmov.u8 r1, q0[11]
+; CHECK-NEXT: vmov.32 q5[3], r1
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vand q5, q5, q2
+; CHECK-NEXT: vmov.u8 r1, q1[0]
+; CHECK-NEXT: vmul.i32 q4, q5, q4
+; CHECK-NEXT: vmov.32 q5[0], r1
+; CHECK-NEXT: vmov.u8 r1, q1[1]
+; CHECK-NEXT: vmov.32 q5[1], r1
+; CHECK-NEXT: vmov.u8 r1, q1[2]
+; CHECK-NEXT: vmov.32 q5[2], r1
+; CHECK-NEXT: vmov.u8 r1, q1[3]
+; CHECK-NEXT: vmov.32 q5[3], r1
+; CHECK-NEXT: vmov.u8 r1, q0[0]
+; CHECK-NEXT: vand q1, q5, q2
+; CHECK-NEXT: vmov.32 q5[0], r1
+; CHECK-NEXT: vmov.u8 r1, q0[1]
+; CHECK-NEXT: vmov.32 q5[1], r1
+; CHECK-NEXT: vmov.u8 r1, q0[2]
+; CHECK-NEXT: vmov.32 q5[2], r1
+; CHECK-NEXT: vmov.u8 r1, q0[3]
+; CHECK-NEXT: vmov.32 q5[3], r1
+; CHECK-NEXT: vand q0, q5, q2
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vadd.i32 q0, q0, q4
+; CHECK-NEXT: vadd.i32 q0, q0, q3
+; CHECK-NEXT: vaddva.u32 r0, q0
+; CHECK-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-NEXT: bx lr
+entry:
+ %xx = zext <16 x i8> %x to <16 x i32>
+ %yy = zext <16 x i8> %y to <16 x i32>
+ %m = mul <16 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v16i32(<16 x i32> %m)
+ %r = add i32 %z, %a
+ ret i32 %r
+}
+
+define arm_aapcs_vfpcc i32 @add_v16i8_v16i32_acc_sext(<16 x i8> %x, <16 x i8> %y, i32 %a) {
+; CHECK-LABEL: add_v16i8_v16i32_acc_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .vsave {d8, d9}
+; CHECK-NEXT: vpush {d8, d9}
+; CHECK-NEXT: vmov.u8 r1, q1[12]
+; CHECK-NEXT: vmov.32 q2[0], r1
+; CHECK-NEXT: vmov.u8 r1, q1[13]
+; CHECK-NEXT: vmov.32 q2[1], r1
+; CHECK-NEXT: vmov.u8 r1, q1[14]
+; CHECK-NEXT: vmov.32 q2[2], r1
+; CHECK-NEXT: vmov.u8 r1, q1[15]
+; CHECK-NEXT: vmov.32 q2[3], r1
+; CHECK-NEXT: vmov.u8 r1, q0[12]
+; CHECK-NEXT: vmov.32 q3[0], r1
+; CHECK-NEXT: vmov.u8 r1, q0[13]
+; CHECK-NEXT: vmov.32 q3[1], r1
+; CHECK-NEXT: vmov.u8 r1, q0[14]
+; CHECK-NEXT: vmov.32 q3[2], r1
+; CHECK-NEXT: vmov.u8 r1, q0[15]
+; CHECK-NEXT: vmov.32 q3[3], r1
+; CHECK-NEXT: vmovlb.s8 q2, q2
+; CHECK-NEXT: vmovlb.s8 q3, q3
+; CHECK-NEXT: vmovlb.s16 q2, q2
+; CHECK-NEXT: vmovlb.s16 q3, q3
+; CHECK-NEXT: vmov.u8 r1, q1[4]
+; CHECK-NEXT: vmul.i32 q2, q3, q2
+; CHECK-NEXT: vmov.32 q3[0], r1
+; CHECK-NEXT: vmov.u8 r1, q1[5]
+; CHECK-NEXT: vmov.32 q3[1], r1
+; CHECK-NEXT: vmov.u8 r1, q1[6]
+; CHECK-NEXT: vmov.32 q3[2], r1
+; CHECK-NEXT: vmov.u8 r1, q1[7]
+; CHECK-NEXT: vmov.32 q3[3], r1
+; CHECK-NEXT: vmov.u8 r1, q0[4]
+; CHECK-NEXT: vmov.32 q4[0], r1
+; CHECK-NEXT: vmov.u8 r1, q0[5]
+; CHECK-NEXT: vmov.32 q4[1], r1
+; CHECK-NEXT: vmov.u8 r1, q0[6]
+; CHECK-NEXT: vmov.32 q4[2], r1
+; CHECK-NEXT: vmov.u8 r1, q0[7]
+; CHECK-NEXT: vmov.32 q4[3], r1
+; CHECK-NEXT: vmovlb.s8 q3, q3
+; CHECK-NEXT: vmovlb.s8 q4, q4
+; CHECK-NEXT: vmovlb.s16 q3, q3
+; CHECK-NEXT: vmovlb.s16 q4, q4
+; CHECK-NEXT: vmov.u8 r1, q1[8]
+; CHECK-NEXT: vmul.i32 q3, q4, q3
+; CHECK-NEXT: vadd.i32 q2, q3, q2
+; CHECK-NEXT: vmov.32 q3[0], r1
+; CHECK-NEXT: vmov.u8 r1, q1[9]
+; CHECK-NEXT: vmov.32 q3[1], r1
+; CHECK-NEXT: vmov.u8 r1, q1[10]
+; CHECK-NEXT: vmov.32 q3[2], r1
+; CHECK-NEXT: vmov.u8 r1, q1[11]
+; CHECK-NEXT: vmov.32 q3[3], r1
+; CHECK-NEXT: vmov.u8 r1, q0[8]
+; CHECK-NEXT: vmov.32 q4[0], r1
+; CHECK-NEXT: vmov.u8 r1, q0[9]
+; CHECK-NEXT: vmov.32 q4[1], r1
+; CHECK-NEXT: vmov.u8 r1, q0[10]
+; CHECK-NEXT: vmov.32 q4[2], r1
+; CHECK-NEXT: vmov.u8 r1, q0[11]
+; CHECK-NEXT: vmov.32 q4[3], r1
+; CHECK-NEXT: vmovlb.s8 q3, q3
+; CHECK-NEXT: vmovlb.s8 q4, q4
+; CHECK-NEXT: vmovlb.s16 q3, q3
+; CHECK-NEXT: vmovlb.s16 q4, q4
+; CHECK-NEXT: vmov.u8 r1, q1[0]
+; CHECK-NEXT: vmul.i32 q3, q4, q3
+; CHECK-NEXT: vmov.32 q4[0], r1
+; CHECK-NEXT: vmov.u8 r1, q1[1]
+; CHECK-NEXT: vmov.32 q4[1], r1
+; CHECK-NEXT: vmov.u8 r1, q1[2]
+; CHECK-NEXT: vmov.32 q4[2], r1
+; CHECK-NEXT: vmov.u8 r1, q1[3]
+; CHECK-NEXT: vmov.32 q4[3], r1
+; CHECK-NEXT: vmov.u8 r1, q0[0]
+; CHECK-NEXT: vmovlb.s8 q1, q4
+; CHECK-NEXT: vmov.32 q4[0], r1
+; CHECK-NEXT: vmov.u8 r1, q0[1]
+; CHECK-NEXT: vmovlb.s16 q1, q1
+; CHECK-NEXT: vmov.32 q4[1], r1
+; CHECK-NEXT: vmov.u8 r1, q0[2]
+; CHECK-NEXT: vmov.32 q4[2], r1
+; CHECK-NEXT: vmov.u8 r1, q0[3]
+; CHECK-NEXT: vmov.32 q4[3], r1
+; CHECK-NEXT: vmovlb.s8 q0, q4
+; CHECK-NEXT: vmovlb.s16 q0, q0
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vadd.i32 q0, q0, q3
+; CHECK-NEXT: vadd.i32 q0, q0, q2
+; CHECK-NEXT: vaddva.u32 r0, q0
+; CHECK-NEXT: vpop {d8, d9}
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <16 x i8> %x to <16 x i32>
+ %yy = sext <16 x i8> %y to <16 x i32>
+ %m = mul <16 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v16i32(<16 x i32> %m)
+ %r = add i32 %z, %a
+ ret i32 %r
+}
+
+define arm_aapcs_vfpcc i32 @add_v4i8_v4i32_acc_zext(<4 x i8> %x, <4 x i8> %y, i32 %a) {
+; CHECK-LABEL: add_v4i8_v4i32_acc_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov.i32 q2, #0xff
+; CHECK-NEXT: vand q1, q1, q2
+; CHECK-NEXT: vand q0, q0, q2
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vaddva.u32 r0, q0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = zext <4 x i8> %x to <4 x i32>
+ %yy = zext <4 x i8> %y to <4 x i32>
+ %m = mul <4 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v4i32(<4 x i32> %m)
+ %r = add i32 %z, %a
+ ret i32 %r
+}
+
+define arm_aapcs_vfpcc i32 @add_v4i8_v4i32_acc_sext(<4 x i8> %x, <4 x i8> %y, i32 %a) {
+; CHECK-LABEL: add_v4i8_v4i32_acc_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmovlb.s8 q1, q1
+; CHECK-NEXT: vmovlb.s8 q0, q0
+; CHECK-NEXT: vmovlb.s16 q1, q1
+; CHECK-NEXT: vmovlb.s16 q0, q0
+; CHECK-NEXT: vmul.i32 q0, q0, q1
+; CHECK-NEXT: vaddva.u32 r0, q0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <4 x i8> %x to <4 x i32>
+ %yy = sext <4 x i8> %y to <4 x i32>
+ %m = mul <4 x i32> %xx, %yy
+ %z = call i32 @llvm.experimental.vector.reduce.add.v4i32(<4 x i32> %m)
+ %r = add i32 %z, %a
+ ret i32 %r
+}
+
+define arm_aapcs_vfpcc zeroext i16 @add_v16i8_v16i16_acc_zext(<16 x i8> %x, <16 x i8> %y, i16 %a) {
+; CHECK-LABEL: add_v16i8_v16i16_acc_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov.u8 r1, q1[8]
+; CHECK-NEXT: vmov.16 q2[0], r1
+; CHECK-NEXT: vmov.u8 r1, q1[9]
+; CHECK-NEXT: vmov.16 q2[1], r1
+; CHECK-NEXT: vmov.u8 r1, q1[10]
+; CHECK-NEXT: vmov.16 q2[2], r1
+; CHECK-NEXT: vmov.u8 r1, q1[11]
+; CHECK-NEXT: vmov.16 q2[3], r1
+; CHECK-NEXT: vmov.u8 r1, q1[12]
+; CHECK-NEXT: vmov.16 q2[4], r1
+; CHECK-NEXT: vmov.u8 r1, q1[13]
+; CHECK-NEXT: vmov.16 q2[5], r1
+; CHECK-NEXT: vmov.u8 r1, q1[14]
+; CHECK-NEXT: vmov.16 q2[6], r1
+; CHECK-NEXT: vmov.u8 r1, q1[15]
+; CHECK-NEXT: vmov.16 q2[7], r1
+; CHECK-NEXT: vmov.u8 r1, q0[8]
+; CHECK-NEXT: vmov.16 q3[0], r1
+; CHECK-NEXT: vmov.u8 r1, q0[9]
+; CHECK-NEXT: vmov.16 q3[1], r1
+; CHECK-NEXT: vmov.u8 r1, q0[10]
+; CHECK-NEXT: vmov.16 q3[2], r1
+; CHECK-NEXT: vmov.u8 r1, q0[11]
+; CHECK-NEXT: vmov.16 q3[3], r1
+; CHECK-NEXT: vmov.u8 r1, q0[12]
+; CHECK-NEXT: vmov.16 q3[4], r1
+; CHECK-NEXT: vmov.u8 r1, q0[13]
+; CHECK-NEXT: vmov.16 q3[5], r1
+; CHECK-NEXT: vmov.u8 r1, q0[14]
+; CHECK-NEXT: vmov.16 q3[6], r1
+; CHECK-NEXT: vmov.u8 r1, q0[15]
+; CHECK-NEXT: vmov.16 q3[7], r1
+; CHECK-NEXT: vmovlb.u8 q2, q2
+; CHECK-NEXT: vmovlb.u8 q3, q3
+; CHECK-NEXT: vmov.u8 r1, q1[0]
+; CHECK-NEXT: vmul.i16 q2, q3, q2
+; CHECK-NEXT: vmov.16 q3[0], r1
+; CHECK-NEXT: vmov.u8 r1, q1[1]
+; CHECK-NEXT: vmov.16 q3[1], r1
+; CHECK-NEXT: vmov.u8 r1, q1[2]
+; CHECK-NEXT: vmov.16 q3[2], r1
+; CHECK-NEXT: vmov.u8 r1, q1[3]
+; CHECK-NEXT: vmov.16 q3[3], r1
+; CHECK-NEXT: vmov.u8 r1, q1[4]
+; CHECK-NEXT: vmov.16 q3[4], r1
+; CHECK-NEXT: vmov.u8 r1, q1[5]
+; CHECK-NEXT: vmov.16 q3[5], r1
+; CHECK-NEXT: vmov.u8 r1, q1[6]
+; CHECK-NEXT: vmov.16 q3[6], r1
+; CHECK-NEXT: vmov.u8 r1, q1[7]
+; CHECK-NEXT: vmov.16 q3[7], r1
+; CHECK-NEXT: vmov.u8 r1, q0[0]
+; CHECK-NEXT: vmovlb.u8 q1, q3
+; CHECK-NEXT: vmov.16 q3[0], r1
+; CHECK-NEXT: vmov.u8 r1, q0[1]
+; CHECK-NEXT: vmov.16 q3[1], r1
+; CHECK-NEXT: vmov.u8 r1, q0[2]
+; CHECK-NEXT: vmov.16 q3[2], r1
+; CHECK-NEXT: vmov.u8 r1, q0[3]
+; CHECK-NEXT: vmov.16 q3[3], r1
+; CHECK-NEXT: vmov.u8 r1, q0[4]
+; CHECK-NEXT: vmov.16 q3[4], r1
+; CHECK-NEXT: vmov.u8 r1, q0[5]
+; CHECK-NEXT: vmov.16 q3[5], r1
+; CHECK-NEXT: vmov.u8 r1, q0[6]
+; CHECK-NEXT: vmov.16 q3[6], r1
+; CHECK-NEXT: vmov.u8 r1, q0[7]
+; CHECK-NEXT: vmov.16 q3[7], r1
+; CHECK-NEXT: vmovlb.u8 q0, q3
+; CHECK-NEXT: vmul.i16 q0, q0, q1
+; CHECK-NEXT: vadd.i16 q0, q0, q2
+; CHECK-NEXT: vaddva.u16 r0, q0
+; CHECK-NEXT: uxth r0, r0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = zext <16 x i8> %x to <16 x i16>
+ %yy = zext <16 x i8> %y to <16 x i16>
+ %m = mul <16 x i16> %xx, %yy
+ %z = call i16 @llvm.experimental.vector.reduce.add.v16i16(<16 x i16> %m)
+ %r = add i16 %z, %a
+ ret i16 %r
+}
+
+define arm_aapcs_vfpcc signext i16 @add_v16i8_v16i16_acc_sext(<16 x i8> %x, <16 x i8> %y, i16 %a) {
+; CHECK-LABEL: add_v16i8_v16i16_acc_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmov.u8 r1, q1[8]
+; CHECK-NEXT: vmov.16 q2[0], r1
+; CHECK-NEXT: vmov.u8 r1, q1[9]
+; CHECK-NEXT: vmov.16 q2[1], r1
+; CHECK-NEXT: vmov.u8 r1, q1[10]
+; CHECK-NEXT: vmov.16 q2[2], r1
+; CHECK-NEXT: vmov.u8 r1, q1[11]
+; CHECK-NEXT: vmov.16 q2[3], r1
+; CHECK-NEXT: vmov.u8 r1, q1[12]
+; CHECK-NEXT: vmov.16 q2[4], r1
+; CHECK-NEXT: vmov.u8 r1, q1[13]
+; CHECK-NEXT: vmov.16 q2[5], r1
+; CHECK-NEXT: vmov.u8 r1, q1[14]
+; CHECK-NEXT: vmov.16 q2[6], r1
+; CHECK-NEXT: vmov.u8 r1, q1[15]
+; CHECK-NEXT: vmov.16 q2[7], r1
+; CHECK-NEXT: vmov.u8 r1, q0[8]
+; CHECK-NEXT: vmov.16 q3[0], r1
+; CHECK-NEXT: vmov.u8 r1, q0[9]
+; CHECK-NEXT: vmov.16 q3[1], r1
+; CHECK-NEXT: vmov.u8 r1, q0[10]
+; CHECK-NEXT: vmov.16 q3[2], r1
+; CHECK-NEXT: vmov.u8 r1, q0[11]
+; CHECK-NEXT: vmov.16 q3[3], r1
+; CHECK-NEXT: vmov.u8 r1, q0[12]
+; CHECK-NEXT: vmov.16 q3[4], r1
+; CHECK-NEXT: vmov.u8 r1, q0[13]
+; CHECK-NEXT: vmov.16 q3[5], r1
+; CHECK-NEXT: vmov.u8 r1, q0[14]
+; CHECK-NEXT: vmov.16 q3[6], r1
+; CHECK-NEXT: vmov.u8 r1, q0[15]
+; CHECK-NEXT: vmov.16 q3[7], r1
+; CHECK-NEXT: vmovlb.s8 q2, q2
+; CHECK-NEXT: vmovlb.s8 q3, q3
+; CHECK-NEXT: vmov.u8 r1, q1[0]
+; CHECK-NEXT: vmul.i16 q2, q3, q2
+; CHECK-NEXT: vmov.16 q3[0], r1
+; CHECK-NEXT: vmov.u8 r1, q1[1]
+; CHECK-NEXT: vmov.16 q3[1], r1
+; CHECK-NEXT: vmov.u8 r1, q1[2]
+; CHECK-NEXT: vmov.16 q3[2], r1
+; CHECK-NEXT: vmov.u8 r1, q1[3]
+; CHECK-NEXT: vmov.16 q3[3], r1
+; CHECK-NEXT: vmov.u8 r1, q1[4]
+; CHECK-NEXT: vmov.16 q3[4], r1
+; CHECK-NEXT: vmov.u8 r1, q1[5]
+; CHECK-NEXT: vmov.16 q3[5], r1
+; CHECK-NEXT: vmov.u8 r1, q1[6]
+; CHECK-NEXT: vmov.16 q3[6], r1
+; CHECK-NEXT: vmov.u8 r1, q1[7]
+; CHECK-NEXT: vmov.16 q3[7], r1
+; CHECK-NEXT: vmov.u8 r1, q0[0]
+; CHECK-NEXT: vmovlb.s8 q1, q3
+; CHECK-NEXT: vmov.16 q3[0], r1
+; CHECK-NEXT: vmov.u8 r1, q0[1]
+; CHECK-NEXT: vmov.16 q3[1], r1
+; CHECK-NEXT: vmov.u8 r1, q0[2]
+; CHECK-NEXT: vmov.16 q3[2], r1
+; CHECK-NEXT: vmov.u8 r1, q0[3]
+; CHECK-NEXT: vmov.16 q3[3], r1
+; CHECK-NEXT: vmov.u8 r1, q0[4]
+; CHECK-NEXT: vmov.16 q3[4], r1
+; CHECK-NEXT: vmov.u8 r1, q0[5]
+; CHECK-NEXT: vmov.16 q3[5], r1
+; CHECK-NEXT: vmov.u8 r1, q0[6]
+; CHECK-NEXT: vmov.16 q3[6], r1
+; CHECK-NEXT: vmov.u8 r1, q0[7]
+; CHECK-NEXT: vmov.16 q3[7], r1
+; CHECK-NEXT: vmovlb.s8 q0, q3
+; CHECK-NEXT: vmul.i16 q0, q0, q1
+; CHECK-NEXT: vadd.i16 q0, q0, q2
+; CHECK-NEXT: vaddva.u16 r0, q0
+; CHECK-NEXT: sxth r0, r0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <16 x i8> %x to <16 x i16>
+ %yy = sext <16 x i8> %y to <16 x i16>
+ %m = mul <16 x i16> %xx, %yy
+ %z = call i16 @llvm.experimental.vector.reduce.add.v16i16(<16 x i16> %m)
+ %r = add i16 %z, %a
+ ret i16 %r
+}
+
+define arm_aapcs_vfpcc zeroext i16 @add_v8i8_v8i16_acc_zext(<8 x i8> %x, <8 x i8> %y, i16 %a) {
+; CHECK-LABEL: add_v8i8_v8i16_acc_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmovlb.u8 q1, q1
+; CHECK-NEXT: vmovlb.u8 q0, q0
+; CHECK-NEXT: vmul.i16 q0, q0, q1
+; CHECK-NEXT: vaddva.u16 r0, q0
+; CHECK-NEXT: uxth r0, r0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = zext <8 x i8> %x to <8 x i16>
+ %yy = zext <8 x i8> %y to <8 x i16>
+ %m = mul <8 x i16> %xx, %yy
+ %z = call i16 @llvm.experimental.vector.reduce.add.v8i16(<8 x i16> %m)
+ %r = add i16 %z, %a
+ ret i16 %r
+}
+
+define arm_aapcs_vfpcc signext i16 @add_v8i8_v8i16_acc_sext(<8 x i8> %x, <8 x i8> %y, i16 %a) {
+; CHECK-LABEL: add_v8i8_v8i16_acc_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmovlb.s8 q1, q1
+; CHECK-NEXT: vmovlb.s8 q0, q0
+; CHECK-NEXT: vmul.i16 q0, q0, q1
+; CHECK-NEXT: vaddva.u16 r0, q0
+; CHECK-NEXT: sxth r0, r0
+; CHECK-NEXT: bx lr
+entry:
+ %xx = sext <8 x i8> %x to <8 x i16>
+ %yy = sext <8 x i8> %y to <8 x i16>
+ %m = mul <8 x i16> %xx, %yy
+ %z = call i16 @llvm.experimental.vector.reduce.add.v8i16(<8 x i16> %m)
+ %r = add i16 %z, %a
+ ret i16 %r
+}
+
+define arm_aapcs_vfpcc zeroext i8 @add_v16i8_v16i8_acc(<16 x i8> %x, <16 x i8> %y, i8 %a) {
+; CHECK-LABEL: add_v16i8_v16i8_acc:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: vmul.i8 q0, q0, q1
+; CHECK-NEXT: vaddva.u8 r0, q0
+; CHECK-NEXT: uxtb r0, r0
+; CHECK-NEXT: bx lr
+entry:
+ %m = mul <16 x i8> %x, %y
+ %z = call i8 @llvm.experimental.vector.reduce.add.v16i8(<16 x i8> %m)
+ %r = add i8 %z, %a
+ ret i8 %r
+}
+
+define arm_aapcs_vfpcc i64 @add_v16i8_v16i64_acc_zext(<16 x i8> %x, <16 x i8> %y, i64 %a) {
+; CHECK-LABEL: add_v16i8_v16i64_acc_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r4, r5, r7, lr}
+; CHECK-NEXT: push {r4, r5, r7, lr}
+; CHECK-NEXT: .vsave {d8, d9, d10, d11}
+; CHECK-NEXT: vpush {d8, d9, d10, d11}
+; CHECK-NEXT: vmov.u8 r2, q1[0]
+; CHECK-NEXT: vmov.u8 r3, q0[0]
+; CHECK-NEXT: vmov.32 q3[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[1]
+; CHECK-NEXT: vmov.32 q3[2], r2
+; CHECK-NEXT: adr r2, .LCPI51_0
+; CHECK-NEXT: vldrw.u32 q2, [r2]
+; CHECK-NEXT: vmov.32 q4[0], r3
+; CHECK-NEXT: vmov.u8 r3, q0[1]
+; CHECK-NEXT: vmov.u8 r4, q0[2]
+; CHECK-NEXT: vmov.32 q4[2], r3
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r2, s14
+; CHECK-NEXT: vmov r3, s18
+; CHECK-NEXT: umull r12, lr, r3, r2
+; CHECK-NEXT: vmov r3, s16
+; CHECK-NEXT: vmov r2, s12
+; CHECK-NEXT: vmov.32 q4[0], r4
+; CHECK-NEXT: vmov.u8 r4, q0[3]
+; CHECK-NEXT: vmov.32 q4[2], r4
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r4, s16
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: orr.w lr, lr, r3
+; CHECK-NEXT: vmov.u8 r3, q1[2]
+; CHECK-NEXT: vmov.32 q3[0], r3
+; CHECK-NEXT: vmov.u8 r3, q1[3]
+; CHECK-NEXT: vmov.32 q3[2], r3
+; CHECK-NEXT: add r2, r12
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vmov r3, s12
+; CHECK-NEXT: umull r3, r4, r4, r3
+; CHECK-NEXT: vmov.32 q5[0], r3
+; CHECK-NEXT: vmov r3, s14
+; CHECK-NEXT: vmov.32 q5[1], r4
+; CHECK-NEXT: vmov r4, s18
+; CHECK-NEXT: umull r3, r4, r4, r3
+; CHECK-NEXT: vmov.32 q5[2], r3
+; CHECK-NEXT: vmov.32 q5[3], r4
+; CHECK-NEXT: vmov r3, s20
+; CHECK-NEXT: vmov r5, s21
+; CHECK-NEXT: adds r2, r2, r3
+; CHECK-NEXT: adc.w r3, lr, r5
+; CHECK-NEXT: vmov r5, s22
+; CHECK-NEXT: adds.w r12, r2, r5
+; CHECK-NEXT: vmov.u8 r5, q1[4]
+; CHECK-NEXT: adcs r3, r4
+; CHECK-NEXT: vmov.u8 r4, q0[4]
+; CHECK-NEXT: vmov.32 q3[0], r5
+; CHECK-NEXT: vmov.u8 r5, q1[5]
+; CHECK-NEXT: vmov.32 q4[0], r4
+; CHECK-NEXT: vmov.u8 r4, q0[5]
+; CHECK-NEXT: vmov.32 q3[2], r5
+; CHECK-NEXT: vmov.32 q4[2], r4
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r5, s12
+; CHECK-NEXT: vmov r4, s16
+; CHECK-NEXT: umull r5, r4, r4, r5
+; CHECK-NEXT: vmov.32 q5[0], r5
+; CHECK-NEXT: vmov r5, s14
+; CHECK-NEXT: vmov.32 q5[1], r4
+; CHECK-NEXT: vmov r4, s18
+; CHECK-NEXT: umull r5, r4, r4, r5
+; CHECK-NEXT: vmov.32 q5[2], r5
+; CHECK-NEXT: vmov.32 q5[3], r4
+; CHECK-NEXT: vmov r2, s20
+; CHECK-NEXT: vmov r5, s21
+; CHECK-NEXT: adds.w r2, r2, r12
+; CHECK-NEXT: adcs r3, r5
+; CHECK-NEXT: vmov r5, s22
+; CHECK-NEXT: adds.w r12, r2, r5
+; CHECK-NEXT: vmov.u8 r5, q1[6]
+; CHECK-NEXT: adcs r3, r4
+; CHECK-NEXT: vmov.u8 r4, q0[6]
+; CHECK-NEXT: vmov.32 q3[0], r5
+; CHECK-NEXT: vmov.u8 r5, q1[7]
+; CHECK-NEXT: vmov.32 q4[0], r4
+; CHECK-NEXT: vmov.u8 r4, q0[7]
+; CHECK-NEXT: vmov.32 q3[2], r5
+; CHECK-NEXT: vmov.32 q4[2], r4
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r5, s12
+; CHECK-NEXT: vmov r4, s16
+; CHECK-NEXT: umull r5, r4, r4, r5
+; CHECK-NEXT: vmov.32 q5[0], r5
+; CHECK-NEXT: vmov r5, s14
+; CHECK-NEXT: vmov.32 q5[1], r4
+; CHECK-NEXT: vmov r4, s18
+; CHECK-NEXT: umull r5, r4, r4, r5
+; CHECK-NEXT: vmov.32 q5[2], r5
+; CHECK-NEXT: vmov.32 q5[3], r4
+; CHECK-NEXT: vmov r2, s20
+; CHECK-NEXT: vmov r5, s21
+; CHECK-NEXT: adds.w r2, r2, r12
+; CHECK-NEXT: adcs r3, r5
+; CHECK-NEXT: vmov r5, s22
+; CHECK-NEXT: adds.w r12, r2, r5
+; CHECK-NEXT: vmov.u8 r5, q1[8]
+; CHECK-NEXT: adcs r3, r4
+; CHECK-NEXT: vmov.u8 r4, q0[8]
+; CHECK-NEXT: vmov.32 q3[0], r5
+; CHECK-NEXT: vmov.u8 r5, q1[9]
+; CHECK-NEXT: vmov.32 q4[0], r4
+; CHECK-NEXT: vmov.u8 r4, q0[9]
+; CHECK-NEXT: vmov.32 q3[2], r5
+; CHECK-NEXT: vmov.32 q4[2], r4
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r5, s12
+; CHECK-NEXT: vmov r4, s16
+; CHECK-NEXT: umull r5, r4, r4, r5
+; CHECK-NEXT: vmov.32 q5[0], r5
+; CHECK-NEXT: vmov r5, s14
+; CHECK-NEXT: vmov.32 q5[1], r4
+; CHECK-NEXT: vmov r4, s18
+; CHECK-NEXT: umull r5, r4, r4, r5
+; CHECK-NEXT: vmov.32 q5[2], r5
+; CHECK-NEXT: vmov.32 q5[3], r4
+; CHECK-NEXT: vmov r2, s20
+; CHECK-NEXT: vmov r5, s21
+; CHECK-NEXT: adds.w r2, r2, r12
+; CHECK-NEXT: adcs r3, r5
+; CHECK-NEXT: vmov r5, s22
+; CHECK-NEXT: adds.w r12, r2, r5
+; CHECK-NEXT: vmov.u8 r5, q1[10]
+; CHECK-NEXT: adcs r3, r4
+; CHECK-NEXT: vmov.u8 r4, q0[10]
+; CHECK-NEXT: vmov.32 q3[0], r5
+; CHECK-NEXT: vmov.u8 r5, q1[11]
+; CHECK-NEXT: vmov.32 q4[0], r4
+; CHECK-NEXT: vmov.u8 r4, q0[11]
+; CHECK-NEXT: vmov.32 q3[2], r5
+; CHECK-NEXT: vmov.32 q4[2], r4
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r5, s12
+; CHECK-NEXT: vmov r4, s16
+; CHECK-NEXT: umull r5, r4, r4, r5
+; CHECK-NEXT: vmov.32 q5[0], r5
+; CHECK-NEXT: vmov r5, s14
+; CHECK-NEXT: vmov.32 q5[1], r4
+; CHECK-NEXT: vmov r4, s18
+; CHECK-NEXT: umull r5, r4, r4, r5
+; CHECK-NEXT: vmov.32 q5[2], r5
+; CHECK-NEXT: vmov.32 q5[3], r4
+; CHECK-NEXT: vmov r2, s20
+; CHECK-NEXT: vmov r5, s21
+; CHECK-NEXT: adds.w r2, r2, r12
+; CHECK-NEXT: adcs r3, r5
+; CHECK-NEXT: vmov r5, s22
+; CHECK-NEXT: adds.w r12, r2, r5
+; CHECK-NEXT: vmov.u8 r5, q1[12]
+; CHECK-NEXT: adcs r3, r4
+; CHECK-NEXT: vmov.u8 r4, q0[12]
+; CHECK-NEXT: vmov.32 q3[0], r5
+; CHECK-NEXT: vmov.u8 r5, q1[13]
+; CHECK-NEXT: vmov.32 q4[0], r4
+; CHECK-NEXT: vmov.u8 r4, q0[13]
+; CHECK-NEXT: vmov.32 q3[2], r5
+; CHECK-NEXT: vmov.32 q4[2], r4
+; CHECK-NEXT: vand q3, q3, q2
+; CHECK-NEXT: vand q4, q4, q2
+; CHECK-NEXT: vmov r5, s12
+; CHECK-NEXT: vmov r4, s16
+; CHECK-NEXT: umull r5, r4, r4, r5
+; CHECK-NEXT: vmov.32 q5[0], r5
+; CHECK-NEXT: vmov r5, s14
+; CHECK-NEXT: vmov.32 q5[1], r4
+; CHECK-NEXT: vmov r4, s18
+; CHECK-NEXT: umull r5, r4, r4, r5
+; CHECK-NEXT: vmov.32 q5[2], r5
+; CHECK-NEXT: vmov.32 q5[3], r4
+; CHECK-NEXT: vmov r2, s20
+; CHECK-NEXT: vmov r5, s21
+; CHECK-NEXT: adds.w r2, r2, r12
+; CHECK-NEXT: adcs r3, r5
+; CHECK-NEXT: vmov r5, s22
+; CHECK-NEXT: adds r2, r2, r5
+; CHECK-NEXT: vmov.u8 r5, q1[14]
+; CHECK-NEXT: vmov.32 q3[0], r5
+; CHECK-NEXT: vmov.u8 r5, q1[15]
+; CHECK-NEXT: adcs r3, r4
+; CHECK-NEXT: vmov.32 q3[2], r5
+; CHECK-NEXT: vmov.u8 r4, q0[14]
+; CHECK-NEXT: vand q1, q3, q2
+; CHECK-NEXT: vmov.32 q3[0], r4
+; CHECK-NEXT: vmov.u8 r4, q0[15]
+; CHECK-NEXT: vmov.32 q3[2], r4
+; CHECK-NEXT: vmov r5, s4
+; CHECK-NEXT: vand q0, q3, q2
+; CHECK-NEXT: vmov r4, s0
+; CHECK-NEXT: umlal r2, r3, r4, r5
+; CHECK-NEXT: vmov r5, s6
+; CHECK-NEXT: vmov r4, s2
+; CHECK-NEXT: umlal r2, r3, r4, r5
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: vpop {d8, d9, d10, d11}
+; CHECK-NEXT: pop {r4, r5, r7, pc}
+; CHECK-NEXT: .p2align 4
+; CHECK-NEXT: @ %bb.1:
+; CHECK-NEXT: .LCPI51_0:
+; CHECK-NEXT: .long 255 @ 0xff
+; CHECK-NEXT: .long 0 @ 0x0
+; CHECK-NEXT: .long 255 @ 0xff
+; CHECK-NEXT: .long 0 @ 0x0
+entry:
+ %xx = zext <16 x i8> %x to <16 x i64>
+ %yy = zext <16 x i8> %y to <16 x i64>
+ %m = mul <16 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v16i64(<16 x i64> %m)
+ %r = add i64 %z, %a
+ ret i64 %r
+}
+
+define arm_aapcs_vfpcc i64 @add_v16i8_v16i64_acc_sext(<16 x i8> %x, <16 x i8> %y, i64 %a) {
+; CHECK-LABEL: add_v16i8_v16i64_acc_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r4, lr}
+; CHECK-NEXT: push {r4, lr}
+; CHECK-NEXT: vmov.u8 r2, q1[0]
+; CHECK-NEXT: vmov.u8 r3, q0[0]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[1]
+; CHECK-NEXT: vmov.32 q2[1], r3
+; CHECK-NEXT: vmov.u8 r3, q0[1]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r3
+; CHECK-NEXT: vmov lr, s10
+; CHECK-NEXT: vmov r2, s8
+; CHECK-NEXT: vmov r12, s9
+; CHECK-NEXT: adds.w lr, lr, r2
+; CHECK-NEXT: vmov.u8 r2, q1[2]
+; CHECK-NEXT: adc.w r12, r12, r3
+; CHECK-NEXT: vmov.u8 r3, q0[2]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[3]
+; CHECK-NEXT: vmov.32 q2[1], r3
+; CHECK-NEXT: vmov.u8 r3, q0[3]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r3, r3, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r3
+; CHECK-NEXT: vmov r4, s8
+; CHECK-NEXT: vmov r2, s9
+; CHECK-NEXT: adds.w r4, r4, lr
+; CHECK-NEXT: adc.w r12, r12, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adds.w lr, r4, r2
+; CHECK-NEXT: vmov.u8 r4, q1[4]
+; CHECK-NEXT: vmov.u8 r2, q0[4]
+; CHECK-NEXT: sxtb r4, r4
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: adc.w r12, r12, r3
+; CHECK-NEXT: smull r2, r4, r2, r4
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[5]
+; CHECK-NEXT: vmov.32 q2[1], r4
+; CHECK-NEXT: vmov.u8 r4, q0[5]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r4, r4
+; CHECK-NEXT: smull r2, r4, r4, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r4
+; CHECK-NEXT: vmov r3, s8
+; CHECK-NEXT: vmov r2, s9
+; CHECK-NEXT: adds.w r3, r3, lr
+; CHECK-NEXT: adc.w r12, r12, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adds.w lr, r3, r2
+; CHECK-NEXT: vmov.u8 r2, q0[6]
+; CHECK-NEXT: adc.w r12, r12, r4
+; CHECK-NEXT: vmov.u8 r4, q1[6]
+; CHECK-NEXT: sxtb r4, r4
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: smull r2, r4, r2, r4
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[7]
+; CHECK-NEXT: vmov.32 q2[1], r4
+; CHECK-NEXT: vmov.u8 r4, q0[7]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r4, r4
+; CHECK-NEXT: smull r2, r4, r4, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r4
+; CHECK-NEXT: vmov r3, s8
+; CHECK-NEXT: vmov r2, s9
+; CHECK-NEXT: adds.w r3, r3, lr
+; CHECK-NEXT: adc.w r12, r12, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adds.w lr, r3, r2
+; CHECK-NEXT: vmov.u8 r2, q0[8]
+; CHECK-NEXT: adc.w r12, r12, r4
+; CHECK-NEXT: vmov.u8 r4, q1[8]
+; CHECK-NEXT: sxtb r4, r4
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: smull r2, r4, r2, r4
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[9]
+; CHECK-NEXT: vmov.32 q2[1], r4
+; CHECK-NEXT: vmov.u8 r4, q0[9]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r4, r4
+; CHECK-NEXT: smull r2, r4, r4, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r4
+; CHECK-NEXT: vmov r3, s8
+; CHECK-NEXT: vmov r2, s9
+; CHECK-NEXT: adds.w r3, r3, lr
+; CHECK-NEXT: adc.w r12, r12, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adds.w lr, r3, r2
+; CHECK-NEXT: vmov.u8 r2, q0[10]
+; CHECK-NEXT: adc.w r12, r12, r4
+; CHECK-NEXT: vmov.u8 r4, q1[10]
+; CHECK-NEXT: sxtb r4, r4
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: smull r2, r4, r2, r4
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[11]
+; CHECK-NEXT: vmov.32 q2[1], r4
+; CHECK-NEXT: vmov.u8 r4, q0[11]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r4, r4
+; CHECK-NEXT: smull r2, r4, r4, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r4
+; CHECK-NEXT: vmov r3, s8
+; CHECK-NEXT: vmov r2, s9
+; CHECK-NEXT: adds.w r3, r3, lr
+; CHECK-NEXT: adc.w r12, r12, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adds.w lr, r3, r2
+; CHECK-NEXT: vmov.u8 r2, q0[12]
+; CHECK-NEXT: adc.w r12, r12, r4
+; CHECK-NEXT: vmov.u8 r4, q1[12]
+; CHECK-NEXT: sxtb r4, r4
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: smull r2, r4, r2, r4
+; CHECK-NEXT: vmov.32 q2[0], r2
+; CHECK-NEXT: vmov.u8 r2, q1[13]
+; CHECK-NEXT: vmov.32 q2[1], r4
+; CHECK-NEXT: vmov.u8 r4, q0[13]
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r4, r4
+; CHECK-NEXT: smull r2, r4, r4, r2
+; CHECK-NEXT: vmov.32 q2[2], r2
+; CHECK-NEXT: vmov.32 q2[3], r4
+; CHECK-NEXT: vmov r3, s8
+; CHECK-NEXT: vmov r2, s9
+; CHECK-NEXT: adds.w r3, r3, lr
+; CHECK-NEXT: adc.w r12, r12, r2
+; CHECK-NEXT: vmov r2, s10
+; CHECK-NEXT: adds r2, r2, r3
+; CHECK-NEXT: adc.w r3, r12, r4
+; CHECK-NEXT: vmov.u8 r4, q1[14]
+; CHECK-NEXT: sxtb.w r12, r4
+; CHECK-NEXT: vmov.u8 r4, q0[14]
+; CHECK-NEXT: sxtb r4, r4
+; CHECK-NEXT: smlal r2, r3, r4, r12
+; CHECK-NEXT: vmov.u8 r4, q1[15]
+; CHECK-NEXT: sxtb.w r12, r4
+; CHECK-NEXT: vmov.u8 r4, q0[15]
+; CHECK-NEXT: sxtb r4, r4
+; CHECK-NEXT: smlal r2, r3, r4, r12
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: pop {r4, pc}
+entry:
+ %xx = sext <16 x i8> %x to <16 x i64>
+ %yy = sext <16 x i8> %y to <16 x i64>
+ %m = mul <16 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v16i64(<16 x i64> %m)
+ %r = add i64 %z, %a
+ ret i64 %r
+}
+
+define arm_aapcs_vfpcc i64 @add_v2i8_v2i64_acc_zext(<2 x i8> %x, <2 x i8> %y, i64 %a) {
+; CHECK-LABEL: add_v2i8_v2i64_acc_zext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r7, lr}
+; CHECK-NEXT: push {r7, lr}
+; CHECK-NEXT: adr r2, .LCPI53_0
+; CHECK-NEXT: vldrw.u32 q2, [r2]
+; CHECK-NEXT: vand q1, q1, q2
+; CHECK-NEXT: vand q0, q0, q2
+; CHECK-NEXT: vmov r2, s6
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: umull r12, lr, r3, r2
+; CHECK-NEXT: vmov r2, s4
+; CHECK-NEXT: vmov r3, s0
+; CHECK-NEXT: umull r2, r3, r3, r2
+; CHECK-NEXT: add r2, r12
+; CHECK-NEXT: orr.w r3, r3, lr
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: adcs r1, r3
+; CHECK-NEXT: pop {r7, pc}
+; CHECK-NEXT: .p2align 4
+; CHECK-NEXT: @ %bb.1:
+; CHECK-NEXT: .LCPI53_0:
+; CHECK-NEXT: .long 255 @ 0xff
+; CHECK-NEXT: .long 0 @ 0x0
+; CHECK-NEXT: .long 255 @ 0xff
+; CHECK-NEXT: .long 0 @ 0x0
+entry:
+ %xx = zext <2 x i8> %x to <2 x i64>
+ %yy = zext <2 x i8> %y to <2 x i64>
+ %m = mul <2 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64> %m)
+ %r = add i64 %z, %a
+ ret i64 %r
+}
+
+define arm_aapcs_vfpcc i64 @add_v2i8_v2i64_acc_sext(<2 x i8> %x, <2 x i8> %y, i64 %a) {
+; CHECK-LABEL: add_v2i8_v2i64_acc_sext:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r7, lr}
+; CHECK-NEXT: push {r7, lr}
+; CHECK-NEXT: vmov r2, s4
+; CHECK-NEXT: vmov r3, s0
+; CHECK-NEXT: sxtb r2, r2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smull r2, r12, r3, r2
+; CHECK-NEXT: vmov r3, s6
+; CHECK-NEXT: sxtb.w lr, r3
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: sxtb r3, r3
+; CHECK-NEXT: smlal r2, r12, r3, lr
+; CHECK-NEXT: adds r0, r0, r2
+; CHECK-NEXT: adc.w r1, r1, r12
+; CHECK-NEXT: pop {r7, pc}
+entry:
+ %xx = sext <2 x i8> %x to <2 x i64>
+ %yy = sext <2 x i8> %y to <2 x i64>
+ %m = mul <2 x i64> %xx, %yy
+ %z = call i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64> %m)
+ %r = add i64 %z, %a
+ ret i64 %r
+}
+
+define arm_aapcs_vfpcc i64 @add_v2i64_v2i64_acc(<2 x i64> %x, <2 x i64> %y, i64 %a) {
+; CHECK-LABEL: add_v2i64_v2i64_acc:
+; CHECK: @ %bb.0: @ %entry
+; CHECK-NEXT: .save {r4, r5, r6, lr}
+; CHECK-NEXT: push {r4, r5, r6, lr}
+; CHECK-NEXT: vmov r2, s4
+; CHECK-NEXT: vmov r3, s0
+; CHECK-NEXT: vmov r4, s5
+; CHECK-NEXT: vmov r6, s7
+; CHECK-NEXT: umull r12, lr, r3, r2
+; CHECK-NEXT: mla r3, r3, r4, lr
+; CHECK-NEXT: vmov r4, s1
+; CHECK-NEXT: vmov.32 q2[0], r12
+; CHECK-NEXT: mla r2, r4, r2, r3
+; CHECK-NEXT: vmov r4, s6
+; CHECK-NEXT: vmov r3, s2
+; CHECK-NEXT: vmov.32 q2[1], r2
+; CHECK-NEXT: vmov r12, s8
+; CHECK-NEXT: umull lr, r5, r3, r4
+; CHECK-NEXT: mla r3, r3, r6, r5
+; CHECK-NEXT: vmov r5, s3
+; CHECK-NEXT: adds.w r6, r12, lr
+; CHECK-NEXT: mla r3, r5, r4, r3
+; CHECK-NEXT: adcs r2, r3
+; CHECK-NEXT: adds r0, r0, r6
+; CHECK-NEXT: adcs r1, r2
+; CHECK-NEXT: pop {r4, r5, r6, pc}
+entry:
+ %m = mul <2 x i64> %x, %y
+ %z = call i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64> %m)
+ %r = add i64 %z, %a
+ ret i64 %r
+}
+
+declare i16 @llvm.experimental.vector.reduce.add.v16i16(<16 x i16>)
+declare i16 @llvm.experimental.vector.reduce.add.v8i16(<8 x i16>)
+declare i32 @llvm.experimental.vector.reduce.add.v16i32(<16 x i32>)
+declare i32 @llvm.experimental.vector.reduce.add.v4i32(<4 x i32>)
+declare i32 @llvm.experimental.vector.reduce.add.v8i32(<8 x i32>)
+declare i64 @llvm.experimental.vector.reduce.add.v16i64(<16 x i64>)
+declare i64 @llvm.experimental.vector.reduce.add.v2i64(<2 x i64>)
+declare i64 @llvm.experimental.vector.reduce.add.v4i64(<4 x i64>)
+declare i64 @llvm.experimental.vector.reduce.add.v8i64(<8 x i64>)
+declare i8 @llvm.experimental.vector.reduce.add.v16i8(<16 x i8>)
More information about the llvm-commits
mailing list