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
132 lines
5.0 KiB
LLVM
132 lines
5.0 KiB
LLVM
; RUN: llc < %s -mtriple=arm64-linux-gnu | FileCheck %s -check-prefixes=CHECK,NOZCM-FPR128-CPU --match-full-lines
|
|
; RUN: llc < %s -mtriple=arm64-apple-macosx -mcpu=generic | FileCheck %s -check-prefixes=CHECK,NOZCM-FPR128-CPU --match-full-lines
|
|
; RUN: llc < %s -mtriple=arm64-apple-macosx -mcpu=apple-m1 | FileCheck %s -check-prefixes=CHECK,ZCM-FPR128-CPU --match-full-lines
|
|
; RUN: llc < %s -mtriple=arm64-apple-macosx -mcpu=apple-m1 -mattr=-zcm-fpr128 | FileCheck %s -check-prefixes=CHECK,NOZCM-FPR128-ATTR --match-full-lines
|
|
; RUN: llc < %s -mtriple=arm64-apple-macosx -mattr=+zcm-fpr128 | FileCheck %s -check-prefixes=CHECK,ZCM-FPR128-ATTR --match-full-lines
|
|
|
|
define void @zero_cycle_regmove_FPR64(double %a, double %b, double %c, double %d) {
|
|
entry:
|
|
; CHECK-LABEL: {{_?zero_cycle_regmove_FPR64}}:{{ *}}{{(;|//)}}{{ *}}@zero_cycle_regmove_FPR64
|
|
|
|
; NOZCM-FPR128-CPU: fmov d0, d2
|
|
; NOZCM-FPR128-CPU: fmov d1, d3
|
|
; NOZCM-FPR128-CPU: fmov [[REG2:d[0-9]+]], d3
|
|
; NOZCM-FPR128-CPU: fmov [[REG1:d[0-9]+]], d2
|
|
; NOZCM-FPR128-CPU-NEXT: bl {{_?foo_double}}
|
|
; NOZCM-FPR128-CPU: fmov d0, [[REG1]]
|
|
; NOZCM-FPR128-CPU: fmov d1, [[REG2]]
|
|
|
|
; ZCM-FPR128-CPU: mov.16b [[REG2:v[0-9]+]], v3
|
|
; ZCM-FPR128-CPU: mov.16b [[REG1:v[0-9]+]], v2
|
|
; ZCM-FPR128-CPU: mov.16b v0, v2
|
|
; ZCM-FPR128-CPU: mov.16b v1, v3
|
|
; ZCM-FPR128-CPU-NEXT: bl {{_?foo_double}}
|
|
; ZCM-FPR128-CPU: mov.16b v0, [[REG1]]
|
|
; ZCM-FPR128-CPU: mov.16b v1, [[REG2]]
|
|
|
|
; NOZCM-FPR128-ATTR: fmov [[REG2:d[0-9]+]], d3
|
|
; NOZCM-FPR128-ATTR: fmov [[REG1:d[0-9]+]], d2
|
|
; NOZCM-FPR128-ATTR: fmov d0, d2
|
|
; NOZCM-FPR128-ATTR: fmov d1, d3
|
|
; NOZCM-FPR128-ATTR-NEXT: bl {{_?foo_double}}
|
|
; NOZCM-FPR128-ATTR: fmov d0, [[REG1]]
|
|
; NOZCM-FPR128-ATTR: fmov d1, [[REG2]]
|
|
|
|
; ZCM-FPR128-ATTR: mov.16b v0, v2
|
|
; ZCM-FPR128-ATTR: mov.16b v1, v3
|
|
; ZCM-FPR128-ATTR: mov.16b [[REG2:v[0-9]+]], v3
|
|
; ZCM-FPR128-ATTR: mov.16b [[REG1:v[0-9]+]], v2
|
|
; ZCM-FPR128-ATTR-NEXT: bl {{_?foo_double}}
|
|
; ZCM-FPR128-ATTR: mov.16b v0, [[REG1]]
|
|
; ZCM-FPR128-ATTR: mov.16b v1, [[REG2]]
|
|
%call = call double @foo_double(double %c, double %d)
|
|
%call1 = call double @foo_double(double %c, double %d)
|
|
unreachable
|
|
}
|
|
|
|
declare float @foo_double(double, double)
|
|
|
|
define void @zero_cycle_regmove_FPR32(float %a, float %b, float %c, float %d) {
|
|
entry:
|
|
; CHECK-LABEL: {{_?zero_cycle_regmove_FPR32}}:{{ *}}{{(;|//)}}{{ *}}@zero_cycle_regmove_FPR32
|
|
|
|
; NOZCM-FPR128-CPU: fmov s0, s2
|
|
; NOZCM-FPR128-CPU: fmov s1, s3
|
|
; NOZCM-FPR128-CPU: fmov [[REG2:s[0-9]+]], s3
|
|
; NOZCM-FPR128-CPU: fmov [[REG1:s[0-9]+]], s2
|
|
; NOZCM-FPR128-CPU-NEXT: bl {{_?foo_float}}
|
|
; NOZCM-FPR128-CPU: fmov s0, [[REG1]]
|
|
; NOZCM-FPR128-CPU: fmov s1, [[REG2]]
|
|
|
|
; ZCM-FPR128-CPU: mov.16b [[REG2:v[0-9]+]], v3
|
|
; ZCM-FPR128-CPU: mov.16b [[REG1:v[0-9]+]], v2
|
|
; ZCM-FPR128-CPU: mov.16b v0, v2
|
|
; ZCM-FPR128-CPU: mov.16b v1, v3
|
|
; ZCM-FPR128-CPU-NEXT: bl {{_?foo_float}}
|
|
; ZCM-FPR128-CPU: mov.16b v0, [[REG1]]
|
|
; ZCM-FPR128-CPU: mov.16b v1, [[REG2]]
|
|
|
|
; NOZCM-FPR128-ATTR: fmov [[REG2:s[0-9]+]], s3
|
|
; NOZCM-FPR128-ATTR: fmov [[REG1:s[0-9]+]], s2
|
|
; NOZCM-FPR128-ATTR: fmov s0, s2
|
|
; NOZCM-FPR128-ATTR: fmov s1, s3
|
|
; NOZCM-FPR128-ATTR-NEXT: bl {{_?foo_float}}
|
|
; NOZCM-FPR128-ATTR: fmov s0, [[REG1]]
|
|
; NOZCM-FPR128-ATTR: fmov s1, [[REG2]]
|
|
|
|
; ZCM-FPR128-ATTR: mov.16b v0, v2
|
|
; ZCM-FPR128-ATTR: mov.16b v1, v3
|
|
; ZCM-FPR128-ATTR: mov.16b [[REG2:v[0-9]+]], v3
|
|
; ZCM-FPR128-ATTR: mov.16b [[REG1:v[0-9]+]], v2
|
|
; ZCM-FPR128-ATTR-NEXT: bl {{_?foo_float}}
|
|
; ZCM-FPR128-ATTR: mov.16b v0, [[REG1]]
|
|
; ZCM-FPR128-ATTR: mov.16b v1, [[REG2]]
|
|
%call = call float @foo_float(float %c, float %d)
|
|
%call1 = call float @foo_float(float %c, float %d)
|
|
unreachable
|
|
}
|
|
|
|
declare float @foo_float(float, float)
|
|
|
|
define void @zero_cycle_regmove_FPR16(half %a, half %b, half %c, half %d) {
|
|
entry:
|
|
; CHECK-LABEL: {{_?zero_cycle_regmove_FPR16}}:{{ *}}{{(;|//)}}{{ *}}@zero_cycle_regmove_FPR16
|
|
|
|
; NOZCM-FPR128-CPU: fmov s0, s2
|
|
; NOZCM-FPR128-CPU: fmov s1, s3
|
|
; NOZCM-FPR128-CPU: fmov [[REG2:s[0-9]+]], s3
|
|
; NOZCM-FPR128-CPU: fmov [[REG1:s[0-9]+]], s2
|
|
; NOZCM-FPR128-CPU-NEXT: bl {{_?foo_half}}
|
|
; NOZCM-FPR128-CPU: fmov s0, [[REG1]]
|
|
; NOZCM-FPR128-CPU: fmov s1, [[REG2]]
|
|
|
|
; ZCM-FPR128-CPU: mov.16b [[REG2:v[0-9]+]], v3
|
|
; ZCM-FPR128-CPU: mov.16b [[REG1:v[0-9]+]], v2
|
|
; ZCM-FPR128-CPU: mov.16b v0, v2
|
|
; ZCM-FPR128-CPU: mov.16b v1, v3
|
|
; ZCM-FPR128-CPU-NEXT: bl {{_?foo_half}}
|
|
; ZCM-FPR128-CPU: mov.16b v0, [[REG1]]
|
|
; ZCM-FPR128-CPU: mov.16b v1, [[REG2]]
|
|
|
|
; NOZCM-FPR128-ATTR: fmov [[REG2:s[0-9]+]], s3
|
|
; NOZCM-FPR128-ATTR: fmov [[REG1:s[0-9]+]], s2
|
|
; NOZCM-FPR128-ATTR: fmov s0, s2
|
|
; NOZCM-FPR128-ATTR: fmov s1, s3
|
|
; NOZCM-FPR128-ATTR-NEXT: bl {{_?foo_half}}
|
|
; NOZCM-FPR128-ATTR: fmov s0, [[REG1]]
|
|
; NOZCM-FPR128-ATTR: fmov s1, [[REG2]]
|
|
|
|
; ZCM-FPR128-ATTR: mov.16b v0, v2
|
|
; ZCM-FPR128-ATTR: mov.16b v1, v3
|
|
; ZCM-FPR128-ATTR: mov.16b [[REG2:v[0-9]+]], v3
|
|
; ZCM-FPR128-ATTR: mov.16b [[REG1:v[0-9]+]], v2
|
|
; ZCM-FPR128-ATTR-NEXT: bl {{_?foo_half}}
|
|
; ZCM-FPR128-ATTR: mov.16b v0, [[REG1]]
|
|
; ZCM-FPR128-ATTR: mov.16b v1, [[REG2]]
|
|
%call = call half @foo_half(half %c, half %d)
|
|
%call1 = call half @foo_half(half %c, half %d)
|
|
unreachable
|
|
}
|
|
|
|
declare half @foo_half(half, half)
|