Files
TheRock/compiler/amd-llvm/llvm/test/CodeGen/LoongArch/lsx/fdiv-reciprocal-estimate.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

122 lines
4.0 KiB
LLVM

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
; RUN: llc --mtriple=loongarch32 --mattr=+32s,+lsx,-frecipe < %s | FileCheck %s --check-prefixes=FAULT,FAULT-LA32
; RUN: llc --mtriple=loongarch32 --mattr=+32s,+lsx,+frecipe < %s | FileCheck %s
; RUN: llc --mtriple=loongarch64 --mattr=+lsx,-frecipe < %s | FileCheck %s --check-prefixes=FAULT,FAULT-LA64
; RUN: llc --mtriple=loongarch64 --mattr=+lsx,+frecipe < %s | FileCheck %s
define void @fdiv_v4f32(ptr %res, ptr %a0, ptr %a1) nounwind {
; FAULT-LABEL: fdiv_v4f32:
; FAULT: # %bb.0: # %entry
; FAULT-NEXT: vld $vr0, $a1, 0
; FAULT-NEXT: vld $vr1, $a2, 0
; FAULT-NEXT: vfdiv.s $vr0, $vr0, $vr1
; FAULT-NEXT: vst $vr0, $a0, 0
; FAULT-NEXT: ret
;
; CHECK-LABEL: fdiv_v4f32:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: vld $vr0, $a2, 0
; CHECK-NEXT: vld $vr1, $a1, 0
; CHECK-NEXT: vfrecipe.s $vr2, $vr0
; CHECK-NEXT: vfmul.s $vr3, $vr1, $vr2
; CHECK-NEXT: vfnmsub.s $vr0, $vr0, $vr3, $vr1
; CHECK-NEXT: vfmadd.s $vr0, $vr2, $vr0, $vr3
; CHECK-NEXT: vst $vr0, $a0, 0
; CHECK-NEXT: ret
entry:
%v0 = load <4 x float>, ptr %a0
%v1 = load <4 x float>, ptr %a1
%v2 = fdiv fast <4 x float> %v0, %v1
store <4 x float> %v2, ptr %res
ret void
}
define void @fdiv_v2f64(ptr %res, ptr %a0, ptr %a1) nounwind {
; FAULT-LABEL: fdiv_v2f64:
; FAULT: # %bb.0: # %entry
; FAULT-NEXT: vld $vr0, $a1, 0
; FAULT-NEXT: vld $vr1, $a2, 0
; FAULT-NEXT: vfdiv.d $vr0, $vr0, $vr1
; FAULT-NEXT: vst $vr0, $a0, 0
; FAULT-NEXT: ret
;
; CHECK-LABEL: fdiv_v2f64:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: vld $vr0, $a2, 0
; CHECK-NEXT: vld $vr1, $a1, 0
; CHECK-NEXT: vfrecipe.d $vr2, $vr0
; CHECK-NEXT: vldi $vr3, -784
; CHECK-NEXT: vfmadd.d $vr3, $vr0, $vr2, $vr3
; CHECK-NEXT: vfnmsub.d $vr2, $vr3, $vr2, $vr2
; CHECK-NEXT: vfmul.d $vr3, $vr1, $vr2
; CHECK-NEXT: vfnmsub.d $vr0, $vr0, $vr3, $vr1
; CHECK-NEXT: vfmadd.d $vr0, $vr2, $vr0, $vr3
; CHECK-NEXT: vst $vr0, $a0, 0
; CHECK-NEXT: ret
entry:
%v0 = load <2 x double>, ptr %a0
%v1 = load <2 x double>, ptr %a1
%v2 = fdiv fast <2 x double> %v0, %v1
store <2 x double> %v2, ptr %res
ret void
}
;; 1.0 / vec
define void @one_fdiv_v4f32(ptr %res, ptr %a0) nounwind {
; FAULT-LABEL: one_fdiv_v4f32:
; FAULT: # %bb.0: # %entry
; FAULT-NEXT: vld $vr0, $a1, 0
; FAULT-NEXT: vfrecip.s $vr0, $vr0
; FAULT-NEXT: vst $vr0, $a0, 0
; FAULT-NEXT: ret
;
; CHECK-LABEL: one_fdiv_v4f32:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: vld $vr0, $a1, 0
; CHECK-NEXT: vfrecipe.s $vr1, $vr0
; CHECK-NEXT: vldi $vr2, -1296
; CHECK-NEXT: vfmadd.s $vr0, $vr0, $vr1, $vr2
; CHECK-NEXT: vfnmsub.s $vr0, $vr0, $vr1, $vr1
; CHECK-NEXT: vst $vr0, $a0, 0
; CHECK-NEXT: ret
entry:
%v0 = load <4 x float>, ptr %a0
%div = fdiv fast <4 x float> <float 1.0, float 1.0, float 1.0, float 1.0>, %v0
store <4 x float> %div, ptr %res
ret void
}
define void @one_fdiv_v2f64(ptr %res, ptr %a0) nounwind {
; FAULT-LA32-LABEL: one_fdiv_v2f64:
; FAULT-LA32: # %bb.0: # %entry
; FAULT-LA32-NEXT: vld $vr0, $a1, 0
; FAULT-LA32-NEXT: vldi $vr1, -912
; FAULT-LA32-NEXT: vfdiv.d $vr0, $vr1, $vr0
; FAULT-LA32-NEXT: vst $vr0, $a0, 0
; FAULT-LA32-NEXT: ret
;
; CHECK-LABEL: one_fdiv_v2f64:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: vld $vr0, $a1, 0
; CHECK-NEXT: vfrecipe.d $vr1, $vr0
; CHECK-NEXT: vldi $vr2, -912
; CHECK-NEXT: vfnmsub.d $vr3, $vr0, $vr1, $vr2
; CHECK-NEXT: vfmadd.d $vr1, $vr1, $vr3, $vr1
; CHECK-NEXT: vfnmsub.d $vr0, $vr0, $vr1, $vr2
; CHECK-NEXT: vfmadd.d $vr0, $vr1, $vr0, $vr1
; CHECK-NEXT: vst $vr0, $a0, 0
; CHECK-NEXT: ret
;
; FAULT-LA64-LABEL: one_fdiv_v2f64:
; FAULT-LA64: # %bb.0: # %entry
; FAULT-LA64-NEXT: vld $vr0, $a1, 0
; FAULT-LA64-NEXT: vfrecip.d $vr0, $vr0
; FAULT-LA64-NEXT: vst $vr0, $a0, 0
; FAULT-LA64-NEXT: ret
entry:
%v0 = load <2 x double>, ptr %a0
%div = fdiv fast <2 x double> <double 1.0, double 1.0>, %v0
store <2 x double> %div, ptr %res
ret void
}