Files
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

353 lines
12 KiB
LLVM

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
; RUN: llc -mtriple=aarch64 -verify-machineinstrs %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-NOFP16,CHECK-NOFP16-SD
; RUN: llc -mtriple=aarch64 -mattr=+fullfp16 -verify-machineinstrs %s -o - | FileCheck %s --check-prefixes=CHECK,CHECK-FP16,CHECK-FP16-SD
; RUN: llc -mtriple=aarch64 -global-isel -global-isel-abort=2 -verify-machineinstrs %s -o - -global-isel-abort=2 | FileCheck %s --check-prefixes=CHECK,CHECK-NOFP16,CHECK-NOFP16-GI
; RUN: llc -mtriple=aarch64 -mattr=+fullfp16 -global-isel -global-isel-abort=2 -verify-machineinstrs %s -o - -global-isel-abort=2 | FileCheck %s --check-prefixes=CHECK,CHECK-FP16,CHECK-FP16-GI
define double @fabs_f64(double %a) {
; CHECK-LABEL: fabs_f64:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: fabs d0, d0
; CHECK-NEXT: ret
entry:
%c = call double @llvm.fabs.f64(double %a)
ret double %c
}
define float @fabs_f32(float %a) {
; CHECK-LABEL: fabs_f32:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: fabs s0, s0
; CHECK-NEXT: ret
entry:
%c = call float @llvm.fabs.f32(float %a)
ret float %c
}
define half @fabs_f16(half %a) {
; CHECK-NOFP16-LABEL: fabs_f16:
; CHECK-NOFP16: // %bb.0: // %entry
; CHECK-NOFP16-NEXT: // kill: def $h0 killed $h0 def $s0
; CHECK-NOFP16-NEXT: fmov w8, s0
; CHECK-NOFP16-NEXT: and w8, w8, #0x7fff
; CHECK-NOFP16-NEXT: fmov s0, w8
; CHECK-NOFP16-NEXT: // kill: def $h0 killed $h0 killed $s0
; CHECK-NOFP16-NEXT: ret
;
; CHECK-FP16-LABEL: fabs_f16:
; CHECK-FP16: // %bb.0: // %entry
; CHECK-FP16-NEXT: fabs h0, h0
; CHECK-FP16-NEXT: ret
entry:
%c = call half @llvm.fabs.f16(half %a)
ret half %c
}
define bfloat @fabs_bf16(bfloat %a) {
; CHECK-NOFP16-LABEL: fabs_bf16:
; CHECK-NOFP16: // %bb.0: // %entry
; CHECK-NOFP16-NEXT: // kill: def $h0 killed $h0 def $s0
; CHECK-NOFP16-NEXT: fmov w8, s0
; CHECK-NOFP16-NEXT: and w8, w8, #0x7fff
; CHECK-NOFP16-NEXT: fmov s0, w8
; CHECK-NOFP16-NEXT: // kill: def $h0 killed $h0 killed $s0
; CHECK-NOFP16-NEXT: ret
;
; CHECK-FP16-LABEL: fabs_bf16:
; CHECK-FP16: // %bb.0: // %entry
; CHECK-FP16-NEXT: fabs h0, h0
; CHECK-FP16-NEXT: ret
entry:
%c = call bfloat @llvm.fabs.bf16(bfloat %a)
ret bfloat %c
}
define <2 x double> @fabs_v2f64(<2 x double> %a) {
; CHECK-LABEL: fabs_v2f64:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: fabs v0.2d, v0.2d
; CHECK-NEXT: ret
entry:
%c = call <2 x double> @llvm.fabs.v2f64(<2 x double> %a)
ret <2 x double> %c
}
define <3 x double> @fabs_v3f64(<3 x double> %a) {
; CHECK-NOFP16-SD-LABEL: fabs_v3f64:
; CHECK-NOFP16-SD: // %bb.0: // %entry
; CHECK-NOFP16-SD-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-NOFP16-SD-NEXT: // kill: def $d1 killed $d1 def $q1
; CHECK-NOFP16-SD-NEXT: // kill: def $d2 killed $d2 def $q2
; CHECK-NOFP16-SD-NEXT: mov v0.d[1], v1.d[0]
; CHECK-NOFP16-SD-NEXT: fabs v2.2d, v2.2d
; CHECK-NOFP16-SD-NEXT: // kill: def $d2 killed $d2 killed $q2
; CHECK-NOFP16-SD-NEXT: fabs v0.2d, v0.2d
; CHECK-NOFP16-SD-NEXT: mov d1, v0.d[1]
; CHECK-NOFP16-SD-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NOFP16-SD-NEXT: ret
;
; CHECK-FP16-SD-LABEL: fabs_v3f64:
; CHECK-FP16-SD: // %bb.0: // %entry
; CHECK-FP16-SD-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-FP16-SD-NEXT: // kill: def $d1 killed $d1 def $q1
; CHECK-FP16-SD-NEXT: // kill: def $d2 killed $d2 def $q2
; CHECK-FP16-SD-NEXT: mov v0.d[1], v1.d[0]
; CHECK-FP16-SD-NEXT: fabs v2.2d, v2.2d
; CHECK-FP16-SD-NEXT: // kill: def $d2 killed $d2 killed $q2
; CHECK-FP16-SD-NEXT: fabs v0.2d, v0.2d
; CHECK-FP16-SD-NEXT: mov d1, v0.d[1]
; CHECK-FP16-SD-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-FP16-SD-NEXT: ret
;
; CHECK-NOFP16-GI-LABEL: fabs_v3f64:
; CHECK-NOFP16-GI: // %bb.0: // %entry
; CHECK-NOFP16-GI-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-NOFP16-GI-NEXT: // kill: def $d1 killed $d1 def $q1
; CHECK-NOFP16-GI-NEXT: fabs d2, d2
; CHECK-NOFP16-GI-NEXT: mov v0.d[1], v1.d[0]
; CHECK-NOFP16-GI-NEXT: fabs v0.2d, v0.2d
; CHECK-NOFP16-GI-NEXT: mov d1, v0.d[1]
; CHECK-NOFP16-GI-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NOFP16-GI-NEXT: ret
;
; CHECK-FP16-GI-LABEL: fabs_v3f64:
; CHECK-FP16-GI: // %bb.0: // %entry
; CHECK-FP16-GI-NEXT: // kill: def $d0 killed $d0 def $q0
; CHECK-FP16-GI-NEXT: // kill: def $d1 killed $d1 def $q1
; CHECK-FP16-GI-NEXT: fabs d2, d2
; CHECK-FP16-GI-NEXT: mov v0.d[1], v1.d[0]
; CHECK-FP16-GI-NEXT: fabs v0.2d, v0.2d
; CHECK-FP16-GI-NEXT: mov d1, v0.d[1]
; CHECK-FP16-GI-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-FP16-GI-NEXT: ret
entry:
%c = call <3 x double> @llvm.fabs.v3f64(<3 x double> %a)
ret <3 x double> %c
}
define <4 x double> @fabs_v4f64(<4 x double> %a) {
; CHECK-LABEL: fabs_v4f64:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: fabs v0.2d, v0.2d
; CHECK-NEXT: fabs v1.2d, v1.2d
; CHECK-NEXT: ret
entry:
%c = call <4 x double> @llvm.fabs.v4f64(<4 x double> %a)
ret <4 x double> %c
}
define <2 x float> @fabs_v2f32(<2 x float> %a) {
; CHECK-LABEL: fabs_v2f32:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: fabs v0.2s, v0.2s
; CHECK-NEXT: ret
entry:
%c = call <2 x float> @llvm.fabs.v2f32(<2 x float> %a)
ret <2 x float> %c
}
define <3 x float> @fabs_v3f32(<3 x float> %a) {
; CHECK-LABEL: fabs_v3f32:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: fabs v0.4s, v0.4s
; CHECK-NEXT: ret
entry:
%c = call <3 x float> @llvm.fabs.v3f32(<3 x float> %a)
ret <3 x float> %c
}
define <4 x float> @fabs_v4f32(<4 x float> %a) {
; CHECK-LABEL: fabs_v4f32:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: fabs v0.4s, v0.4s
; CHECK-NEXT: ret
entry:
%c = call <4 x float> @llvm.fabs.v4f32(<4 x float> %a)
ret <4 x float> %c
}
define <8 x float> @fabs_v8f32(<8 x float> %a) {
; CHECK-LABEL: fabs_v8f32:
; CHECK: // %bb.0: // %entry
; CHECK-NEXT: fabs v0.4s, v0.4s
; CHECK-NEXT: fabs v1.4s, v1.4s
; CHECK-NEXT: ret
entry:
%c = call <8 x float> @llvm.fabs.v8f32(<8 x float> %a)
ret <8 x float> %c
}
define <7 x half> @fabs_v7f16(<7 x half> %a) {
; CHECK-NOFP16-SD-LABEL: fabs_v7f16:
; CHECK-NOFP16-SD: // %bb.0: // %entry
; CHECK-NOFP16-SD-NEXT: bic v0.8h, #128, lsl #8
; CHECK-NOFP16-SD-NEXT: ret
;
; CHECK-FP16-LABEL: fabs_v7f16:
; CHECK-FP16: // %bb.0: // %entry
; CHECK-FP16-NEXT: fabs v0.8h, v0.8h
; CHECK-FP16-NEXT: ret
;
; CHECK-NOFP16-GI-LABEL: fabs_v7f16:
; CHECK-NOFP16-GI: // %bb.0: // %entry
; CHECK-NOFP16-GI-NEXT: mvni v1.8h, #128, lsl #8
; CHECK-NOFP16-GI-NEXT: and v0.16b, v0.16b, v1.16b
; CHECK-NOFP16-GI-NEXT: ret
entry:
%c = call <7 x half> @llvm.fabs.v7f16(<7 x half> %a)
ret <7 x half> %c
}
define <4 x half> @fabs_v4f16(<4 x half> %a) {
; CHECK-NOFP16-SD-LABEL: fabs_v4f16:
; CHECK-NOFP16-SD: // %bb.0: // %entry
; CHECK-NOFP16-SD-NEXT: bic v0.4h, #128, lsl #8
; CHECK-NOFP16-SD-NEXT: ret
;
; CHECK-FP16-LABEL: fabs_v4f16:
; CHECK-FP16: // %bb.0: // %entry
; CHECK-FP16-NEXT: fabs v0.4h, v0.4h
; CHECK-FP16-NEXT: ret
;
; CHECK-NOFP16-GI-LABEL: fabs_v4f16:
; CHECK-NOFP16-GI: // %bb.0: // %entry
; CHECK-NOFP16-GI-NEXT: mvni v1.4h, #128, lsl #8
; CHECK-NOFP16-GI-NEXT: and v0.8b, v0.8b, v1.8b
; CHECK-NOFP16-GI-NEXT: ret
entry:
%c = call <4 x half> @llvm.fabs.v4f16(<4 x half> %a)
ret <4 x half> %c
}
define <8 x half> @fabs_v8f16(<8 x half> %a) {
; CHECK-NOFP16-SD-LABEL: fabs_v8f16:
; CHECK-NOFP16-SD: // %bb.0: // %entry
; CHECK-NOFP16-SD-NEXT: bic v0.8h, #128, lsl #8
; CHECK-NOFP16-SD-NEXT: ret
;
; CHECK-FP16-LABEL: fabs_v8f16:
; CHECK-FP16: // %bb.0: // %entry
; CHECK-FP16-NEXT: fabs v0.8h, v0.8h
; CHECK-FP16-NEXT: ret
;
; CHECK-NOFP16-GI-LABEL: fabs_v8f16:
; CHECK-NOFP16-GI: // %bb.0: // %entry
; CHECK-NOFP16-GI-NEXT: mvni v1.8h, #128, lsl #8
; CHECK-NOFP16-GI-NEXT: and v0.16b, v0.16b, v1.16b
; CHECK-NOFP16-GI-NEXT: ret
entry:
%c = call <8 x half> @llvm.fabs.v8f16(<8 x half> %a)
ret <8 x half> %c
}
define <16 x half> @fabs_v16f16(<16 x half> %a) {
; CHECK-NOFP16-SD-LABEL: fabs_v16f16:
; CHECK-NOFP16-SD: // %bb.0: // %entry
; CHECK-NOFP16-SD-NEXT: bic v0.8h, #128, lsl #8
; CHECK-NOFP16-SD-NEXT: bic v1.8h, #128, lsl #8
; CHECK-NOFP16-SD-NEXT: ret
;
; CHECK-FP16-LABEL: fabs_v16f16:
; CHECK-FP16: // %bb.0: // %entry
; CHECK-FP16-NEXT: fabs v0.8h, v0.8h
; CHECK-FP16-NEXT: fabs v1.8h, v1.8h
; CHECK-FP16-NEXT: ret
;
; CHECK-NOFP16-GI-LABEL: fabs_v16f16:
; CHECK-NOFP16-GI: // %bb.0: // %entry
; CHECK-NOFP16-GI-NEXT: mvni v2.8h, #128, lsl #8
; CHECK-NOFP16-GI-NEXT: and v0.16b, v0.16b, v2.16b
; CHECK-NOFP16-GI-NEXT: and v1.16b, v1.16b, v2.16b
; CHECK-NOFP16-GI-NEXT: ret
entry:
%c = call <16 x half> @llvm.fabs.v16f16(<16 x half> %a)
ret <16 x half> %c
}
define <7 x bfloat> @fabs_v7bf16(<7 x bfloat> %a) {
; CHECK-NOFP16-SD-LABEL: fabs_v7bf16:
; CHECK-NOFP16-SD: // %bb.0: // %entry
; CHECK-NOFP16-SD-NEXT: bic v0.8h, #128, lsl #8
; CHECK-NOFP16-SD-NEXT: ret
;
; CHECK-FP16-LABEL: fabs_v7bf16:
; CHECK-FP16: // %bb.0: // %entry
; CHECK-FP16-NEXT: fabs v0.8h, v0.8h
; CHECK-FP16-NEXT: ret
;
; CHECK-NOFP16-GI-LABEL: fabs_v7bf16:
; CHECK-NOFP16-GI: // %bb.0: // %entry
; CHECK-NOFP16-GI-NEXT: mvni v1.8h, #128, lsl #8
; CHECK-NOFP16-GI-NEXT: and v0.16b, v0.16b, v1.16b
; CHECK-NOFP16-GI-NEXT: ret
entry:
%c = call <7 x bfloat> @llvm.fabs.v7bf16(<7 x bfloat> %a)
ret <7 x bfloat> %c
}
define <4 x bfloat> @fabs_v4bf16(<4 x bfloat> %a) {
; CHECK-NOFP16-SD-LABEL: fabs_v4bf16:
; CHECK-NOFP16-SD: // %bb.0: // %entry
; CHECK-NOFP16-SD-NEXT: bic v0.4h, #128, lsl #8
; CHECK-NOFP16-SD-NEXT: ret
;
; CHECK-FP16-LABEL: fabs_v4bf16:
; CHECK-FP16: // %bb.0: // %entry
; CHECK-FP16-NEXT: fabs v0.4h, v0.4h
; CHECK-FP16-NEXT: ret
;
; CHECK-NOFP16-GI-LABEL: fabs_v4bf16:
; CHECK-NOFP16-GI: // %bb.0: // %entry
; CHECK-NOFP16-GI-NEXT: mvni v1.4h, #128, lsl #8
; CHECK-NOFP16-GI-NEXT: and v0.8b, v0.8b, v1.8b
; CHECK-NOFP16-GI-NEXT: ret
entry:
%c = call <4 x bfloat> @llvm.fabs.v4bf16(<4 x bfloat> %a)
ret <4 x bfloat> %c
}
define <8 x bfloat> @fabs_v8bf16(<8 x bfloat> %a) {
; CHECK-NOFP16-SD-LABEL: fabs_v8bf16:
; CHECK-NOFP16-SD: // %bb.0: // %entry
; CHECK-NOFP16-SD-NEXT: bic v0.8h, #128, lsl #8
; CHECK-NOFP16-SD-NEXT: ret
;
; CHECK-FP16-LABEL: fabs_v8bf16:
; CHECK-FP16: // %bb.0: // %entry
; CHECK-FP16-NEXT: fabs v0.8h, v0.8h
; CHECK-FP16-NEXT: ret
;
; CHECK-NOFP16-GI-LABEL: fabs_v8bf16:
; CHECK-NOFP16-GI: // %bb.0: // %entry
; CHECK-NOFP16-GI-NEXT: mvni v1.8h, #128, lsl #8
; CHECK-NOFP16-GI-NEXT: and v0.16b, v0.16b, v1.16b
; CHECK-NOFP16-GI-NEXT: ret
entry:
%c = call <8 x bfloat> @llvm.fabs.v8bf16(<8 x bfloat> %a)
ret <8 x bfloat> %c
}
define <16 x bfloat> @fabs_v16bf16(<16 x bfloat> %a) {
; CHECK-NOFP16-SD-LABEL: fabs_v16bf16:
; CHECK-NOFP16-SD: // %bb.0: // %entry
; CHECK-NOFP16-SD-NEXT: bic v0.8h, #128, lsl #8
; CHECK-NOFP16-SD-NEXT: bic v1.8h, #128, lsl #8
; CHECK-NOFP16-SD-NEXT: ret
;
; CHECK-FP16-LABEL: fabs_v16bf16:
; CHECK-FP16: // %bb.0: // %entry
; CHECK-FP16-NEXT: fabs v0.8h, v0.8h
; CHECK-FP16-NEXT: fabs v1.8h, v1.8h
; CHECK-FP16-NEXT: ret
;
; CHECK-NOFP16-GI-LABEL: fabs_v16bf16:
; CHECK-NOFP16-GI: // %bb.0: // %entry
; CHECK-NOFP16-GI-NEXT: mvni v2.8h, #128, lsl #8
; CHECK-NOFP16-GI-NEXT: and v0.16b, v0.16b, v2.16b
; CHECK-NOFP16-GI-NEXT: and v1.16b, v1.16b, v2.16b
; CHECK-NOFP16-GI-NEXT: ret
entry:
%c = call <16 x bfloat> @llvm.fabs.v16bf16(<16 x bfloat> %a)
ret <16 x bfloat> %c
}