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
113 lines
7.6 KiB
C
113 lines
7.6 KiB
C
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 6
|
|
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +neon -target-feature +f16f32dot -disable-O0-optnone -emit-llvm -o - %s | opt -S -passes=mem2reg,sroa,instcombine | FileCheck %s
|
|
// RUN: %clang_cc1 -x c++ -triple aarch64-none-linux-gnu -target-feature +neon -target-feature +f16f32dot -disable-O0-optnone -emit-llvm -o - %s | opt -S -passes=mem2reg,sroa,instcombine | FileCheck %s -check-prefix CHECK-CXX
|
|
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +neon -target-feature +f16f32dot -O3 -S -o /dev/null %s
|
|
|
|
// REQUIRES: aarch64-registered-target
|
|
|
|
#include <arm_neon.h>
|
|
|
|
// CHECK-LABEL: define dso_local <2 x float> @test_vdot_f32_f16(
|
|
// CHECK-SAME: <2 x float> noundef [[R:%.*]], <4 x half> noundef [[A:%.*]], <4 x half> noundef [[B:%.*]]) #[[ATTR0:[0-9]+]] {
|
|
// CHECK-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-NEXT: [[VDOT3_I:%.*]] = call <2 x float> @llvm.aarch64.neon.fdot.v2f32.v4f16(<2 x float> [[R]], <4 x half> [[A]], <4 x half> [[B]])
|
|
// CHECK-NEXT: ret <2 x float> [[VDOT3_I]]
|
|
//
|
|
// CHECK-CXX-LABEL: define dso_local noundef <2 x float> @_Z17test_vdot_f32_f1613__Float32x2_t13__Float16x4_tS0_(
|
|
// CHECK-CXX-SAME: <2 x float> noundef [[R:%.*]], <4 x half> noundef [[A:%.*]], <4 x half> noundef [[B:%.*]]) #[[ATTR0:[0-9]+]] {
|
|
// CHECK-CXX-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-CXX-NEXT: [[VDOT3_I:%.*]] = call <2 x float> @llvm.aarch64.neon.fdot.v2f32.v4f16(<2 x float> [[R]], <4 x half> [[A]], <4 x half> [[B]])
|
|
// CHECK-CXX-NEXT: ret <2 x float> [[VDOT3_I]]
|
|
//
|
|
float32x2_t test_vdot_f32_f16(float32x2_t r, float16x4_t a, float16x4_t b) {
|
|
return vdot_f32_f16(r, a, b);
|
|
}
|
|
|
|
// CHECK-LABEL: define dso_local <4 x float> @test_vdotq_f32_f16(
|
|
// CHECK-SAME: <4 x float> noundef [[R:%.*]], <8 x half> noundef [[A:%.*]], <8 x half> noundef [[B:%.*]]) #[[ATTR0]] {
|
|
// CHECK-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-NEXT: [[VDOT3_I:%.*]] = call <4 x float> @llvm.aarch64.neon.fdot.v4f32.v8f16(<4 x float> [[R]], <8 x half> [[A]], <8 x half> [[B]])
|
|
// CHECK-NEXT: ret <4 x float> [[VDOT3_I]]
|
|
//
|
|
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z18test_vdotq_f32_f1613__Float32x4_t13__Float16x8_tS0_(
|
|
// CHECK-CXX-SAME: <4 x float> noundef [[R:%.*]], <8 x half> noundef [[A:%.*]], <8 x half> noundef [[B:%.*]]) #[[ATTR0]] {
|
|
// CHECK-CXX-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-CXX-NEXT: [[VDOT3_I:%.*]] = call <4 x float> @llvm.aarch64.neon.fdot.v4f32.v8f16(<4 x float> [[R]], <8 x half> [[A]], <8 x half> [[B]])
|
|
// CHECK-CXX-NEXT: ret <4 x float> [[VDOT3_I]]
|
|
//
|
|
float32x4_t test_vdotq_f32_f16(float32x4_t r, float16x8_t a, float16x8_t b) {
|
|
return vdotq_f32_f16(r, a, b);
|
|
}
|
|
|
|
// CHECK-LABEL: define dso_local <2 x float> @test_vdot_lane_f32_f16(
|
|
// CHECK-SAME: <2 x float> noundef [[R:%.*]], <4 x half> noundef [[A:%.*]], <4 x half> noundef [[B:%.*]]) #[[ATTR0]] {
|
|
// CHECK-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-NEXT: [[LANE:%.*]] = shufflevector <4 x half> [[B]], <4 x half> poison, <4 x i32> zeroinitializer
|
|
// CHECK-NEXT: [[VDOT2:%.*]] = call <2 x float> @llvm.aarch64.neon.fdot.v2f32.v4f16(<2 x float> [[R]], <4 x half> [[A]], <4 x half> [[LANE]])
|
|
// CHECK-NEXT: ret <2 x float> [[VDOT2]]
|
|
//
|
|
// CHECK-CXX-LABEL: define dso_local noundef <2 x float> @_Z22test_vdot_lane_f32_f1613__Float32x2_t13__Float16x4_tS0_(
|
|
// CHECK-CXX-SAME: <2 x float> noundef [[R:%.*]], <4 x half> noundef [[A:%.*]], <4 x half> noundef [[B:%.*]]) #[[ATTR0]] {
|
|
// CHECK-CXX-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-CXX-NEXT: [[LANE:%.*]] = shufflevector <4 x half> [[B]], <4 x half> poison, <4 x i32> zeroinitializer
|
|
// CHECK-CXX-NEXT: [[VDOT2:%.*]] = call <2 x float> @llvm.aarch64.neon.fdot.v2f32.v4f16(<2 x float> [[R]], <4 x half> [[A]], <4 x half> [[LANE]])
|
|
// CHECK-CXX-NEXT: ret <2 x float> [[VDOT2]]
|
|
//
|
|
float32x2_t test_vdot_lane_f32_f16(float32x2_t r, float16x4_t a, float16x4_t b) {
|
|
return vdot_lane_f32_f16(r, a, b, 0);
|
|
}
|
|
|
|
// CHECK-LABEL: define dso_local <2 x float> @test_vdot_laneq_f32_f16(
|
|
// CHECK-SAME: <2 x float> noundef [[R:%.*]], <4 x half> noundef [[A:%.*]], <8 x half> noundef [[B:%.*]]) #[[ATTR0]] {
|
|
// CHECK-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-NEXT: [[LANE:%.*]] = shufflevector <8 x half> [[B]], <8 x half> poison, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
|
|
// CHECK-NEXT: [[VDOT2:%.*]] = call <2 x float> @llvm.aarch64.neon.fdot.v2f32.v4f16(<2 x float> [[R]], <4 x half> [[A]], <4 x half> [[LANE]])
|
|
// CHECK-NEXT: ret <2 x float> [[VDOT2]]
|
|
//
|
|
// CHECK-CXX-LABEL: define dso_local noundef <2 x float> @_Z23test_vdot_laneq_f32_f1613__Float32x2_t13__Float16x4_t13__Float16x8_t(
|
|
// CHECK-CXX-SAME: <2 x float> noundef [[R:%.*]], <4 x half> noundef [[A:%.*]], <8 x half> noundef [[B:%.*]]) #[[ATTR0]] {
|
|
// CHECK-CXX-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-CXX-NEXT: [[LANE:%.*]] = shufflevector <8 x half> [[B]], <8 x half> poison, <4 x i32> <i32 3, i32 3, i32 3, i32 3>
|
|
// CHECK-CXX-NEXT: [[VDOT2:%.*]] = call <2 x float> @llvm.aarch64.neon.fdot.v2f32.v4f16(<2 x float> [[R]], <4 x half> [[A]], <4 x half> [[LANE]])
|
|
// CHECK-CXX-NEXT: ret <2 x float> [[VDOT2]]
|
|
//
|
|
float32x2_t test_vdot_laneq_f32_f16(float32x2_t r, float16x4_t a, float16x8_t b) {
|
|
return vdot_laneq_f32_f16(r, a, b, 3);
|
|
}
|
|
|
|
// CHECK-LABEL: define dso_local <4 x float> @test_vdotq_lane_f32_f16(
|
|
// CHECK-SAME: <4 x float> noundef [[R:%.*]], <8 x half> noundef [[A:%.*]], <4 x half> noundef [[B:%.*]]) #[[ATTR0]] {
|
|
// CHECK-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-NEXT: [[LANE:%.*]] = shufflevector <4 x half> [[B]], <4 x half> poison, <8 x i32> zeroinitializer
|
|
// CHECK-NEXT: [[VDOT2:%.*]] = call <4 x float> @llvm.aarch64.neon.fdot.v4f32.v8f16(<4 x float> [[R]], <8 x half> [[A]], <8 x half> [[LANE]])
|
|
// CHECK-NEXT: ret <4 x float> [[VDOT2]]
|
|
//
|
|
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z23test_vdotq_lane_f32_f1613__Float32x4_t13__Float16x8_t13__Float16x4_t(
|
|
// CHECK-CXX-SAME: <4 x float> noundef [[R:%.*]], <8 x half> noundef [[A:%.*]], <4 x half> noundef [[B:%.*]]) #[[ATTR0]] {
|
|
// CHECK-CXX-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-CXX-NEXT: [[LANE:%.*]] = shufflevector <4 x half> [[B]], <4 x half> poison, <8 x i32> zeroinitializer
|
|
// CHECK-CXX-NEXT: [[VDOT2:%.*]] = call <4 x float> @llvm.aarch64.neon.fdot.v4f32.v8f16(<4 x float> [[R]], <8 x half> [[A]], <8 x half> [[LANE]])
|
|
// CHECK-CXX-NEXT: ret <4 x float> [[VDOT2]]
|
|
//
|
|
float32x4_t test_vdotq_lane_f32_f16(float32x4_t r, float16x8_t a, float16x4_t b) {
|
|
return vdotq_lane_f32_f16(r, a, b, 0);
|
|
}
|
|
|
|
// CHECK-LABEL: define dso_local <4 x float> @test_vdotq_laneq_f32_f16(
|
|
// CHECK-SAME: <4 x float> noundef [[R:%.*]], <8 x half> noundef [[A:%.*]], <8 x half> noundef [[B:%.*]]) #[[ATTR0]] {
|
|
// CHECK-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-NEXT: [[LANE:%.*]] = shufflevector <8 x half> [[B]], <8 x half> poison, <8 x i32> <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
|
|
// CHECK-NEXT: [[VDOT2:%.*]] = call <4 x float> @llvm.aarch64.neon.fdot.v4f32.v8f16(<4 x float> [[R]], <8 x half> [[A]], <8 x half> [[LANE]])
|
|
// CHECK-NEXT: ret <4 x float> [[VDOT2]]
|
|
//
|
|
// CHECK-CXX-LABEL: define dso_local noundef <4 x float> @_Z24test_vdotq_laneq_f32_f1613__Float32x4_t13__Float16x8_tS0_(
|
|
// CHECK-CXX-SAME: <4 x float> noundef [[R:%.*]], <8 x half> noundef [[A:%.*]], <8 x half> noundef [[B:%.*]]) #[[ATTR0]] {
|
|
// CHECK-CXX-NEXT: [[ENTRY:.*:]]
|
|
// CHECK-CXX-NEXT: [[LANE:%.*]] = shufflevector <8 x half> [[B]], <8 x half> poison, <8 x i32> <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
|
|
// CHECK-CXX-NEXT: [[VDOT2:%.*]] = call <4 x float> @llvm.aarch64.neon.fdot.v4f32.v8f16(<4 x float> [[R]], <8 x half> [[A]], <8 x half> [[LANE]])
|
|
// CHECK-CXX-NEXT: ret <4 x float> [[VDOT2]]
|
|
//
|
|
float32x4_t test_vdotq_laneq_f32_f16(float32x4_t r, float16x8_t a, float16x8_t b) {
|
|
return vdotq_laneq_f32_f16(r, a, b, 3);
|
|
}
|