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
161 lines
6.0 KiB
LLVM
161 lines
6.0 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=aarch64-- | FileCheck %s
|
|
|
|
; Each pair of tests should be logically equivalent,
|
|
; so codegen should be the same and optimal for each pair.
|
|
|
|
define <4 x float> @splat0_before_fmul_constant(<4 x float> %a) {
|
|
; CHECK-LABEL: splat0_before_fmul_constant:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: fmov v1.4s, #3.00000000
|
|
; CHECK-NEXT: fmul v0.4s, v1.4s, v0.s[0]
|
|
; CHECK-NEXT: ret
|
|
%splat = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> zeroinitializer
|
|
%mul = fmul <4 x float> %splat, <float 3.0, float 3.0, float 3.0, float 3.0>
|
|
ret <4 x float> %mul
|
|
}
|
|
|
|
define <4 x float> @splat0_after_fmul_constant(<4 x float> %a) {
|
|
; CHECK-LABEL: splat0_after_fmul_constant:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: fmov v1.4s, #3.00000000
|
|
; CHECK-NEXT: fmul v0.4s, v0.4s, v1.4s
|
|
; CHECK-NEXT: dup v0.4s, v0.s[0]
|
|
; CHECK-NEXT: ret
|
|
%mul = fmul <4 x float> %a, <float 3.0, float 42.0, float 3.0, float 3.0>
|
|
%splat = shufflevector <4 x float> %mul, <4 x float> undef, <4 x i32> zeroinitializer
|
|
ret <4 x float> %splat
|
|
}
|
|
|
|
; Try different type and splat lane.
|
|
|
|
define <2 x double> @splat1_before_fmul_constant(<2 x double> %a) {
|
|
; CHECK-LABEL: splat1_before_fmul_constant:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: fmov v1.2d, #5.00000000
|
|
; CHECK-NEXT: fmul v0.2d, v1.2d, v0.d[1]
|
|
; CHECK-NEXT: ret
|
|
%splat = shufflevector <2 x double> %a, <2 x double> undef, <2 x i32> <i32 1, i32 1>
|
|
%mul = fmul <2 x double> %splat, <double 5.0, double 5.0>
|
|
ret <2 x double> %mul
|
|
}
|
|
|
|
define <2 x double> @splat1_after_fmul_constant(<2 x double> %a) {
|
|
; CHECK-LABEL: splat1_after_fmul_constant:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: fmov v1.2d, #5.00000000
|
|
; CHECK-NEXT: fmul v0.2d, v0.2d, v1.2d
|
|
; CHECK-NEXT: dup v0.2d, v0.d[1]
|
|
; CHECK-NEXT: ret
|
|
%mul = fmul <2 x double> %a, <double -1.0, double 5.0>
|
|
%splat = shufflevector <2 x double> %mul, <2 x double> undef, <2 x i32> <i32 1, i32 1>
|
|
ret <2 x double> %splat
|
|
}
|
|
|
|
; 2 variable operands
|
|
|
|
define <2 x double> @splat1_before_fmul(<2 x double> %a, <2 x double> %b) {
|
|
; CHECK-LABEL: splat1_before_fmul:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: fmul v0.2d, v0.2d, v1.2d
|
|
; CHECK-NEXT: dup v0.2d, v0.d[1]
|
|
; CHECK-NEXT: ret
|
|
%splata = shufflevector <2 x double> %a, <2 x double> undef, <2 x i32> <i32 1, i32 1>
|
|
%splatb = shufflevector <2 x double> %b, <2 x double> undef, <2 x i32> <i32 1, i32 1>
|
|
%mul = fmul <2 x double> %splata, %splatb
|
|
ret <2 x double> %mul
|
|
}
|
|
|
|
define <2 x double> @splat1_after_fmul(<2 x double> %a, <2 x double> %b) {
|
|
; CHECK-LABEL: splat1_after_fmul:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: fmul v0.2d, v0.2d, v1.2d
|
|
; CHECK-NEXT: dup v0.2d, v0.d[1]
|
|
; CHECK-NEXT: ret
|
|
%mul = fmul <2 x double> %a, %b
|
|
%splat = shufflevector <2 x double> %mul, <2 x double> undef, <2 x i32> <i32 1, i32 1>
|
|
ret <2 x double> %splat
|
|
}
|
|
|
|
; Integer multiply
|
|
|
|
define <4 x i32> @splat2_before_mul_constant(<4 x i32> %a) {
|
|
; CHECK-LABEL: splat2_before_mul_constant:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: movi v1.4s, #3
|
|
; CHECK-NEXT: mul v0.4s, v1.4s, v0.s[2]
|
|
; CHECK-NEXT: ret
|
|
%splat = shufflevector <4 x i32> %a, <4 x i32> undef, <4 x i32> <i32 2, i32 undef, i32 2, i32 undef>
|
|
%mul = mul <4 x i32> %splat, <i32 3, i32 3, i32 3, i32 3>
|
|
ret <4 x i32> %mul
|
|
}
|
|
|
|
define <4 x i32> @splat2_after_mul_constant(<4 x i32> %a) {
|
|
; CHECK-LABEL: splat2_after_mul_constant:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: movi v1.4s, #3
|
|
; CHECK-NEXT: mul v0.4s, v0.4s, v1.4s
|
|
; CHECK-NEXT: dup v0.4s, v0.s[2]
|
|
; CHECK-NEXT: ret
|
|
%mul = mul <4 x i32> %a, <i32 3, i32 3, i32 3, i32 3>
|
|
%splat = shufflevector <4 x i32> %mul, <4 x i32> undef, <4 x i32> <i32 2, i32 undef, i32 2, i32 undef>
|
|
ret <4 x i32> %splat
|
|
}
|
|
|
|
; Different type, lane, and 2 variable operands.
|
|
|
|
define <8 x i16> @splat1_before_mul(<8 x i16> %a, <8 x i16> %b) {
|
|
; CHECK-LABEL: splat1_before_mul:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mul v0.8h, v0.8h, v1.8h
|
|
; CHECK-NEXT: dup v0.8h, v0.h[1]
|
|
; CHECK-NEXT: ret
|
|
%splata = shufflevector <8 x i16> %a, <8 x i16> undef, <8 x i32> <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
|
|
%splatb = shufflevector <8 x i16> %b, <8 x i16> undef, <8 x i32> <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
|
|
%mul = mul <8 x i16> %splata, %splatb
|
|
ret <8 x i16> %mul
|
|
}
|
|
|
|
define <8 x i16> @splat1_after_mul(<8 x i16> %a, <8 x i16> %b) {
|
|
; CHECK-LABEL: splat1_after_mul:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: mul v0.8h, v0.8h, v1.8h
|
|
; CHECK-NEXT: dup v0.8h, v0.h[1]
|
|
; CHECK-NEXT: ret
|
|
%mul = mul <8 x i16> %a, %b
|
|
%splat = shufflevector <8 x i16> %mul, <8 x i16> undef, <8 x i32> <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
|
|
ret <8 x i16> %splat
|
|
}
|
|
|
|
; Multiple multiplies.
|
|
|
|
define <4 x float> @splat0_before_fmul_fmul_constant(<4 x float> %a) {
|
|
; CHECK-LABEL: splat0_before_fmul_fmul_constant:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: fmov v1.4s, #3.00000000
|
|
; CHECK-NEXT: fmul v0.4s, v0.4s, v1.4s
|
|
; CHECK-NEXT: fmov v1.4s, #6.00000000
|
|
; CHECK-NEXT: fmul v0.4s, v1.4s, v0.s[0]
|
|
; CHECK-NEXT: ret
|
|
%splat1 = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> zeroinitializer
|
|
%mul1 = fmul <4 x float> %splat1, <float 3.0, float 3.0, float 3.0, float 3.0>
|
|
%splat2 = shufflevector <4 x float> %mul1, <4 x float> undef, <4 x i32> zeroinitializer
|
|
%mul2 = fmul <4 x float> %splat2, <float 6.0, float 6.0, float 6.0, float 6.0>
|
|
ret <4 x float> %mul2
|
|
}
|
|
|
|
define <4 x float> @splat0_after_fmul_fmul_constant(<4 x float> %a) {
|
|
; CHECK-LABEL: splat0_after_fmul_fmul_constant:
|
|
; CHECK: // %bb.0:
|
|
; CHECK-NEXT: fmov v1.4s, #3.00000000
|
|
; CHECK-NEXT: fmul v0.4s, v0.4s, v1.4s
|
|
; CHECK-NEXT: fmov v1.4s, #6.00000000
|
|
; CHECK-NEXT: fmul v0.4s, v0.4s, v1.4s
|
|
; CHECK-NEXT: dup v0.4s, v0.s[0]
|
|
; CHECK-NEXT: ret
|
|
%mul1 = fmul <4 x float> %a, <float 3.0, float 42.0, float 3.0, float 3.0>
|
|
%mul2 = fmul <4 x float> %mul1, <float 6.0, float 42.0, float 3.0, float 3.0>
|
|
%splat = shufflevector <4 x float> %mul2, <4 x float> undef, <4 x i32> zeroinitializer
|
|
ret <4 x float> %splat
|
|
}
|