Files
TheRock/compiler/amd-llvm/llvm/test/CodeGen/AArch64/st1-lane.ll
T
derek.holloway 7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
Flatten subprojects into a single base monorepo
2026-09-02 16:34:47 -07:00

423 lines
15 KiB
LLVM

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; RUN: llc < %s -mtriple=aarch64-none-eabi | FileCheck %s --check-prefixes=CHECK,CHECK-SD
; RUN: llc < %s -mtriple=aarch64-none-eabi -global-isel | FileCheck %s --check-prefixes=CHECK,CHECK-GI
define void @v2i64(<2 x i64> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: v2i64:
; CHECK: // %bb.0:
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: st1 { v0.d }[1], [x1]
; CHECK-NEXT: ret
%s1 = shufflevector <2 x i64> %a, <2 x i64> poison, <1 x i32> <i32 0>
store <1 x i64> %s1, ptr %p1, align 8
%s2 = shufflevector <2 x i64> %a, <2 x i64> poison, <1 x i32> <i32 1>
store <1 x i64> %s2, ptr %p2, align 8
ret void
}
define void @v4i32(<4 x i32> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: v4i32:
; CHECK: // %bb.0:
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: st1 { v0.d }[1], [x1]
; CHECK-NEXT: ret
%s1 = shufflevector <4 x i32> %a, <4 x i32> poison, <2 x i32> <i32 0, i32 1>
store <2 x i32> %s1, ptr %p1, align 8
%s2 = shufflevector <4 x i32> %a, <4 x i32> poison, <2 x i32> <i32 2, i32 3>
store <2 x i32> %s2, ptr %p2, align 8
ret void
}
define void @v8i16(<8 x i16> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: v8i16:
; CHECK: // %bb.0:
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: st1 { v0.d }[1], [x1]
; CHECK-NEXT: ret
%s1 = shufflevector <8 x i16> %a, <8 x i16> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
store <4 x i16> %s1, ptr %p1, align 8
%s2 = shufflevector <8 x i16> %a, <8 x i16> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
store <4 x i16> %s2, ptr %p2, align 8
ret void
}
define void @v16i8(<16 x i8> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: v16i8:
; CHECK: // %bb.0:
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: st1 { v0.d }[1], [x1]
; CHECK-NEXT: ret
%s1 = shufflevector <16 x i8> %a, <16 x i8> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
store <8 x i8> %s1, ptr %p1, align 8
%s2 = shufflevector <16 x i8> %a, <16 x i8> poison, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
store <8 x i8> %s2, ptr %p2, align 8
ret void
}
define void @v2f64(<2 x double> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: v2f64:
; CHECK: // %bb.0:
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: st1 { v0.d }[1], [x1]
; CHECK-NEXT: ret
%s1 = shufflevector <2 x double> %a, <2 x double> poison, <1 x i32> <i32 0>
store <1 x double> %s1, ptr %p1, align 8
%s2 = shufflevector <2 x double> %a, <2 x double> poison, <1 x i32> <i32 1>
store <1 x double> %s2, ptr %p2, align 8
ret void
}
define void @v4f32(<4 x float> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: v4f32:
; CHECK: // %bb.0:
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: st1 { v0.d }[1], [x1]
; CHECK-NEXT: ret
%s1 = shufflevector <4 x float> %a, <4 x float> poison, <2 x i32> <i32 0, i32 1>
store <2 x float> %s1, ptr %p1, align 8
%s2 = shufflevector <4 x float> %a, <4 x float> poison, <2 x i32> <i32 2, i32 3>
store <2 x float> %s2, ptr %p2, align 8
ret void
}
define void @v8f16(<8 x half> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: v8f16:
; CHECK: // %bb.0:
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: st1 { v0.d }[1], [x1]
; CHECK-NEXT: ret
%s1 = shufflevector <8 x half> %a, <8 x half> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
store <4 x half> %s1, ptr %p1, align 8
%s2 = shufflevector <8 x half> %a, <8 x half> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
store <4 x half> %s2, ptr %p2, align 8
ret void
}
define void @v8bf16(<8 x bfloat> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: v8bf16:
; CHECK: // %bb.0:
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: st1 { v0.d }[1], [x1]
; CHECK-NEXT: ret
%s1 = shufflevector <8 x bfloat> %a, <8 x bfloat> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
store <4 x bfloat> %s1, ptr %p1, align 8
%s2 = shufflevector <8 x bfloat> %a, <8 x bfloat> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
store <4 x bfloat> %s2, ptr %p2, align 8
ret void
}
define ptr @post_v2i64(<2 x i64> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: post_v2i64:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: str d1, [x1], #8
; CHECK-NEXT: mov x0, x1
; CHECK-NEXT: ret
%s1 = shufflevector <2 x i64> %a, <2 x i64> poison, <1 x i32> <i32 0>
store <1 x i64> %s1, ptr %p1, align 8
%s2 = shufflevector <2 x i64> %a, <2 x i64> poison, <1 x i32> <i32 1>
store <1 x i64> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 8
ret ptr %p3
}
define ptr @post_v4i32(<4 x i32> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: post_v4i32:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: str d1, [x1], #8
; CHECK-NEXT: mov x0, x1
; CHECK-NEXT: ret
%s1 = shufflevector <4 x i32> %a, <4 x i32> poison, <2 x i32> <i32 0, i32 1>
store <2 x i32> %s1, ptr %p1, align 8
%s2 = shufflevector <4 x i32> %a, <4 x i32> poison, <2 x i32> <i32 2, i32 3>
store <2 x i32> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 8
ret ptr %p3
}
define ptr @post_v8i16(<8 x i16> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: post_v8i16:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: str d1, [x1], #8
; CHECK-NEXT: mov x0, x1
; CHECK-NEXT: ret
%s1 = shufflevector <8 x i16> %a, <8 x i16> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
store <4 x i16> %s1, ptr %p1, align 8
%s2 = shufflevector <8 x i16> %a, <8 x i16> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
store <4 x i16> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 8
ret ptr %p3
}
define ptr @post_v16i8(<16 x i8> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: post_v16i8:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: str d1, [x1], #8
; CHECK-NEXT: mov x0, x1
; CHECK-NEXT: ret
%s1 = shufflevector <16 x i8> %a, <16 x i8> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
store <8 x i8> %s1, ptr %p1, align 8
%s2 = shufflevector <16 x i8> %a, <16 x i8> poison, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
store <8 x i8> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 8
ret ptr %p3
}
define ptr @post_v2f64(<2 x double> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: post_v2f64:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: str d1, [x1], #8
; CHECK-NEXT: mov x0, x1
; CHECK-NEXT: ret
%s1 = shufflevector <2 x double> %a, <2 x double> poison, <1 x i32> <i32 0>
store <1 x double> %s1, ptr %p1, align 8
%s2 = shufflevector <2 x double> %a, <2 x double> poison, <1 x i32> <i32 1>
store <1 x double> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 8
ret ptr %p3
}
define ptr @post_v4f32(<4 x float> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: post_v4f32:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: str d1, [x1], #8
; CHECK-NEXT: mov x0, x1
; CHECK-NEXT: ret
%s1 = shufflevector <4 x float> %a, <4 x float> poison, <2 x i32> <i32 0, i32 1>
store <2 x float> %s1, ptr %p1, align 8
%s2 = shufflevector <4 x float> %a, <4 x float> poison, <2 x i32> <i32 2, i32 3>
store <2 x float> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 8
ret ptr %p3
}
define ptr @post_v8f16(<8 x half> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: post_v8f16:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: str d1, [x1], #8
; CHECK-NEXT: mov x0, x1
; CHECK-NEXT: ret
%s1 = shufflevector <8 x half> %a, <8 x half> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
store <4 x half> %s1, ptr %p1, align 8
%s2 = shufflevector <8 x half> %a, <8 x half> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
store <4 x half> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 8
ret ptr %p3
}
define ptr @post_v8bf16(<8 x bfloat> %a, ptr %p1, ptr %p2) {
; CHECK-LABEL: post_v8bf16:
; CHECK: // %bb.0:
; CHECK-NEXT: mov d1, v0.d[1]
; CHECK-NEXT: str d0, [x0]
; CHECK-NEXT: str d1, [x1], #8
; CHECK-NEXT: mov x0, x1
; CHECK-NEXT: ret
%s1 = shufflevector <8 x bfloat> %a, <8 x bfloat> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
store <4 x bfloat> %s1, ptr %p1, align 8
%s2 = shufflevector <8 x bfloat> %a, <8 x bfloat> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
store <4 x bfloat> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 8
ret ptr %p3
}
define ptr @postx_v2i64(<2 x i64> %a, ptr %p1, ptr %p2, i64 %o) {
; CHECK-SD-LABEL: postx_v2i64:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: mov x8, x0
; CHECK-SD-NEXT: add x0, x1, x2
; CHECK-SD-NEXT: str d0, [x8]
; CHECK-SD-NEXT: st1 { v0.d }[1], [x1]
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: postx_v2i64:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: str d0, [x0]
; CHECK-GI-NEXT: add x0, x1, x2
; CHECK-GI-NEXT: st1 { v0.d }[1], [x1]
; CHECK-GI-NEXT: ret
%s1 = shufflevector <2 x i64> %a, <2 x i64> poison, <1 x i32> <i32 0>
store <1 x i64> %s1, ptr %p1, align 8
%s2 = shufflevector <2 x i64> %a, <2 x i64> poison, <1 x i32> <i32 1>
store <1 x i64> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 %o
ret ptr %p3
}
define ptr @postx_v4i32(<4 x i32> %a, ptr %p1, ptr %p2, i64 %o) {
; CHECK-SD-LABEL: postx_v4i32:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: mov x8, x0
; CHECK-SD-NEXT: add x0, x1, x2
; CHECK-SD-NEXT: str d0, [x8]
; CHECK-SD-NEXT: st1 { v0.d }[1], [x1]
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: postx_v4i32:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: str d0, [x0]
; CHECK-GI-NEXT: add x0, x1, x2
; CHECK-GI-NEXT: st1 { v0.d }[1], [x1]
; CHECK-GI-NEXT: ret
%s1 = shufflevector <4 x i32> %a, <4 x i32> poison, <2 x i32> <i32 0, i32 1>
store <2 x i32> %s1, ptr %p1, align 8
%s2 = shufflevector <4 x i32> %a, <4 x i32> poison, <2 x i32> <i32 2, i32 3>
store <2 x i32> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 %o
ret ptr %p3
}
define ptr @postx_v8i16(<8 x i16> %a, ptr %p1, ptr %p2, i64 %o) {
; CHECK-SD-LABEL: postx_v8i16:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: mov x8, x0
; CHECK-SD-NEXT: add x0, x1, x2
; CHECK-SD-NEXT: str d0, [x8]
; CHECK-SD-NEXT: st1 { v0.d }[1], [x1]
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: postx_v8i16:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: str d0, [x0]
; CHECK-GI-NEXT: add x0, x1, x2
; CHECK-GI-NEXT: st1 { v0.d }[1], [x1]
; CHECK-GI-NEXT: ret
%s1 = shufflevector <8 x i16> %a, <8 x i16> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
store <4 x i16> %s1, ptr %p1, align 8
%s2 = shufflevector <8 x i16> %a, <8 x i16> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
store <4 x i16> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 %o
ret ptr %p3
}
define ptr @postx_v16i8(<16 x i8> %a, ptr %p1, ptr %p2, i64 %o) {
; CHECK-SD-LABEL: postx_v16i8:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: mov x8, x0
; CHECK-SD-NEXT: add x0, x1, x2
; CHECK-SD-NEXT: str d0, [x8]
; CHECK-SD-NEXT: st1 { v0.d }[1], [x1]
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: postx_v16i8:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: str d0, [x0]
; CHECK-GI-NEXT: add x0, x1, x2
; CHECK-GI-NEXT: st1 { v0.d }[1], [x1]
; CHECK-GI-NEXT: ret
%s1 = shufflevector <16 x i8> %a, <16 x i8> poison, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
store <8 x i8> %s1, ptr %p1, align 8
%s2 = shufflevector <16 x i8> %a, <16 x i8> poison, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
store <8 x i8> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 %o
ret ptr %p3
}
define ptr @postx_v2f64(<2 x double> %a, ptr %p1, ptr %p2, i64 %o) {
; CHECK-SD-LABEL: postx_v2f64:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: mov x8, x0
; CHECK-SD-NEXT: add x0, x1, x2
; CHECK-SD-NEXT: str d0, [x8]
; CHECK-SD-NEXT: st1 { v0.d }[1], [x1]
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: postx_v2f64:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: str d0, [x0]
; CHECK-GI-NEXT: add x0, x1, x2
; CHECK-GI-NEXT: st1 { v0.d }[1], [x1]
; CHECK-GI-NEXT: ret
%s1 = shufflevector <2 x double> %a, <2 x double> poison, <1 x i32> <i32 0>
store <1 x double> %s1, ptr %p1, align 8
%s2 = shufflevector <2 x double> %a, <2 x double> poison, <1 x i32> <i32 1>
store <1 x double> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 %o
ret ptr %p3
}
define ptr @postx_v4f32(<4 x float> %a, ptr %p1, ptr %p2, i64 %o) {
; CHECK-SD-LABEL: postx_v4f32:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: mov x8, x0
; CHECK-SD-NEXT: add x0, x1, x2
; CHECK-SD-NEXT: str d0, [x8]
; CHECK-SD-NEXT: st1 { v0.d }[1], [x1]
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: postx_v4f32:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: str d0, [x0]
; CHECK-GI-NEXT: add x0, x1, x2
; CHECK-GI-NEXT: st1 { v0.d }[1], [x1]
; CHECK-GI-NEXT: ret
%s1 = shufflevector <4 x float> %a, <4 x float> poison, <2 x i32> <i32 0, i32 1>
store <2 x float> %s1, ptr %p1, align 8
%s2 = shufflevector <4 x float> %a, <4 x float> poison, <2 x i32> <i32 2, i32 3>
store <2 x float> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 %o
ret ptr %p3
}
define ptr @postx_v8f16(<8 x half> %a, ptr %p1, ptr %p2, i64 %o) {
; CHECK-SD-LABEL: postx_v8f16:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: mov x8, x0
; CHECK-SD-NEXT: add x0, x1, x2
; CHECK-SD-NEXT: str d0, [x8]
; CHECK-SD-NEXT: st1 { v0.d }[1], [x1]
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: postx_v8f16:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: str d0, [x0]
; CHECK-GI-NEXT: add x0, x1, x2
; CHECK-GI-NEXT: st1 { v0.d }[1], [x1]
; CHECK-GI-NEXT: ret
%s1 = shufflevector <8 x half> %a, <8 x half> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
store <4 x half> %s1, ptr %p1, align 8
%s2 = shufflevector <8 x half> %a, <8 x half> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
store <4 x half> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 %o
ret ptr %p3
}
define ptr @postx_v8bf16(<8 x bfloat> %a, ptr %p1, ptr %p2, i64 %o) {
; CHECK-SD-LABEL: postx_v8bf16:
; CHECK-SD: // %bb.0:
; CHECK-SD-NEXT: mov x8, x0
; CHECK-SD-NEXT: add x0, x1, x2
; CHECK-SD-NEXT: str d0, [x8]
; CHECK-SD-NEXT: st1 { v0.d }[1], [x1]
; CHECK-SD-NEXT: ret
;
; CHECK-GI-LABEL: postx_v8bf16:
; CHECK-GI: // %bb.0:
; CHECK-GI-NEXT: str d0, [x0]
; CHECK-GI-NEXT: add x0, x1, x2
; CHECK-GI-NEXT: st1 { v0.d }[1], [x1]
; CHECK-GI-NEXT: ret
%s1 = shufflevector <8 x bfloat> %a, <8 x bfloat> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
store <4 x bfloat> %s1, ptr %p1, align 8
%s2 = shufflevector <8 x bfloat> %a, <8 x bfloat> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
store <4 x bfloat> %s2, ptr %p2, align 8
%p3 = getelementptr i8, ptr %p2, i64 %o
ret ptr %p3
}