Files
TheRock/compiler/amd-llvm/llvm/test/CodeGen/X86/llc-accept-avx10-512.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

98 lines
4.3 KiB
LLVM

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; avx10.x-512 is just avx10.x -- 512 is kept for compatibility purposes.
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx10.1-512 2>&1 | FileCheck --check-prefixes=CHECK-AVX10_1 %s
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx10.2-512 2>&1 | FileCheck --check-prefixes=CHECK-AVX10_2 %s
; CHECK-AVX10_1-NOT: is not recognizable
; CHECK-AVX10_2-NOT: is not recognizable
define <32 x bfloat> @foo_avx10.1(<16 x float> %a, <16 x float> %b) {
; CHECK-AVX10_1-LABEL: foo_avx10.1:
; CHECK-AVX10_1: # %bb.0:
; CHECK-AVX10_1-NEXT: vcvtne2ps2bf16 %zmm1, %zmm0, %zmm0
; CHECK-AVX10_1-NEXT: retq
;
; CHECK-AVX10_2-LABEL: foo_avx10.1:
; CHECK-AVX10_2: # %bb.0:
; CHECK-AVX10_2-NEXT: vcvtne2ps2bf16 %zmm1, %zmm0, %zmm0
; CHECK-AVX10_2-NEXT: retq
%ret = call <32 x bfloat> @llvm.x86.avx512bf16.cvtne2ps2bf16.512(<16 x float> %a, <16 x float> %b)
ret <32 x bfloat> %ret
}
define <8 x i32> @foo_avx10.2(<8 x double> %f) {
; CHECK-AVX10_1-LABEL: foo_avx10.2:
; CHECK-AVX10_1: # %bb.0:
; CHECK-AVX10_1-NEXT: vextractf32x4 $2, %zmm0, %xmm1
; CHECK-AVX10_1-NEXT: vshufpd {{.*#+}} xmm2 = xmm1[1,0]
; CHECK-AVX10_1-NEXT: vmovsd {{.*#+}} xmm3 = [-2.147483648E+9,0.0E+0]
; CHECK-AVX10_1-NEXT: vmaxsd %xmm3, %xmm2, %xmm4
; CHECK-AVX10_1-NEXT: vmovsd {{.*#+}} xmm5 = [2.147483647E+9,0.0E+0]
; CHECK-AVX10_1-NEXT: vminsd %xmm5, %xmm4, %xmm4
; CHECK-AVX10_1-NEXT: vcvttsd2si %xmm4, %ecx
; CHECK-AVX10_1-NEXT: xorl %eax, %eax
; CHECK-AVX10_1-NEXT: vucomisd %xmm2, %xmm2
; CHECK-AVX10_1-NEXT: cmovpl %eax, %ecx
; CHECK-AVX10_1-NEXT: vmaxsd %xmm3, %xmm1, %xmm2
; CHECK-AVX10_1-NEXT: vminsd %xmm5, %xmm2, %xmm2
; CHECK-AVX10_1-NEXT: vcvttsd2si %xmm2, %edx
; CHECK-AVX10_1-NEXT: vucomisd %xmm1, %xmm1
; CHECK-AVX10_1-NEXT: cmovpl %eax, %edx
; CHECK-AVX10_1-NEXT: vmovd %edx, %xmm1
; CHECK-AVX10_1-NEXT: vpinsrd $1, %ecx, %xmm1, %xmm1
; CHECK-AVX10_1-NEXT: vextractf32x4 $3, %zmm0, %xmm2
; CHECK-AVX10_1-NEXT: vmaxsd %xmm3, %xmm2, %xmm4
; CHECK-AVX10_1-NEXT: vminsd %xmm5, %xmm4, %xmm4
; CHECK-AVX10_1-NEXT: vcvttsd2si %xmm4, %ecx
; CHECK-AVX10_1-NEXT: vucomisd %xmm2, %xmm2
; CHECK-AVX10_1-NEXT: cmovpl %eax, %ecx
; CHECK-AVX10_1-NEXT: vpinsrd $2, %ecx, %xmm1, %xmm1
; CHECK-AVX10_1-NEXT: vshufpd {{.*#+}} xmm2 = xmm2[1,0]
; CHECK-AVX10_1-NEXT: vmaxsd %xmm3, %xmm2, %xmm4
; CHECK-AVX10_1-NEXT: vminsd %xmm5, %xmm4, %xmm4
; CHECK-AVX10_1-NEXT: vcvttsd2si %xmm4, %ecx
; CHECK-AVX10_1-NEXT: vucomisd %xmm2, %xmm2
; CHECK-AVX10_1-NEXT: cmovpl %eax, %ecx
; CHECK-AVX10_1-NEXT: vpinsrd $3, %ecx, %xmm1, %xmm1
; CHECK-AVX10_1-NEXT: vshufpd {{.*#+}} xmm2 = xmm0[1,0]
; CHECK-AVX10_1-NEXT: vmaxsd %xmm3, %xmm2, %xmm4
; CHECK-AVX10_1-NEXT: vminsd %xmm5, %xmm4, %xmm4
; CHECK-AVX10_1-NEXT: vcvttsd2si %xmm4, %ecx
; CHECK-AVX10_1-NEXT: vucomisd %xmm2, %xmm2
; CHECK-AVX10_1-NEXT: cmovpl %eax, %ecx
; CHECK-AVX10_1-NEXT: vmaxsd %xmm3, %xmm0, %xmm2
; CHECK-AVX10_1-NEXT: vminsd %xmm5, %xmm2, %xmm2
; CHECK-AVX10_1-NEXT: vcvttsd2si %xmm2, %edx
; CHECK-AVX10_1-NEXT: vucomisd %xmm0, %xmm0
; CHECK-AVX10_1-NEXT: cmovpl %eax, %edx
; CHECK-AVX10_1-NEXT: vmovd %edx, %xmm2
; CHECK-AVX10_1-NEXT: vpinsrd $1, %ecx, %xmm2, %xmm2
; CHECK-AVX10_1-NEXT: vextractf128 $1, %ymm0, %xmm0
; CHECK-AVX10_1-NEXT: vmaxsd %xmm3, %xmm0, %xmm4
; CHECK-AVX10_1-NEXT: vminsd %xmm5, %xmm4, %xmm4
; CHECK-AVX10_1-NEXT: vcvttsd2si %xmm4, %ecx
; CHECK-AVX10_1-NEXT: vucomisd %xmm0, %xmm0
; CHECK-AVX10_1-NEXT: cmovpl %eax, %ecx
; CHECK-AVX10_1-NEXT: vpinsrd $2, %ecx, %xmm2, %xmm2
; CHECK-AVX10_1-NEXT: vshufpd {{.*#+}} xmm0 = xmm0[1,0]
; CHECK-AVX10_1-NEXT: vmaxsd %xmm3, %xmm0, %xmm3
; CHECK-AVX10_1-NEXT: vminsd %xmm5, %xmm3, %xmm3
; CHECK-AVX10_1-NEXT: vcvttsd2si %xmm3, %ecx
; CHECK-AVX10_1-NEXT: vucomisd %xmm0, %xmm0
; CHECK-AVX10_1-NEXT: cmovpl %eax, %ecx
; CHECK-AVX10_1-NEXT: vpinsrd $3, %ecx, %xmm2, %xmm0
; CHECK-AVX10_1-NEXT: vinserti128 $1, %xmm1, %ymm0, %ymm0
; CHECK-AVX10_1-NEXT: retq
;
; CHECK-AVX10_2-LABEL: foo_avx10.2:
; CHECK-AVX10_2: # %bb.0:
; CHECK-AVX10_2-NEXT: vcvttpd2dqs %zmm0, %ymm0
; CHECK-AVX10_2-NEXT: retq
%x = call <8 x i32> @llvm.fptosi.sat.v8i32.v8f64(<8 x double> %f)
ret <8 x i32> %x
}