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
96 lines
3.8 KiB
LLVM
96 lines
3.8 KiB
LLVM
; RUN: llc < %s -mtriple=nvptx64 -mcpu=sm_30 -mattr=+ptx60 | FileCheck %s
|
|
; RUN: %if ptxas-isa-6.0 %{ llc < %s -mtriple=nvptx64 -mcpu=sm_30 -mattr=+ptx60 | %ptxas-verify %}
|
|
|
|
declare i32 @llvm.nvvm.shfl.sync.down.i32(i32, i32, i32, i32)
|
|
declare float @llvm.nvvm.shfl.sync.down.f32(float, i32, i32, i32)
|
|
declare i32 @llvm.nvvm.shfl.sync.up.i32(i32, i32, i32, i32)
|
|
declare float @llvm.nvvm.shfl.sync.up.f32(float, i32, i32, i32)
|
|
declare i32 @llvm.nvvm.shfl.sync.bfly.i32(i32, i32, i32, i32)
|
|
declare float @llvm.nvvm.shfl.sync.bfly.f32(float, i32, i32, i32)
|
|
declare i32 @llvm.nvvm.shfl.sync.idx.i32(i32, i32, i32, i32)
|
|
declare float @llvm.nvvm.shfl.sync.idx.f32(float, i32, i32, i32)
|
|
|
|
; CHECK-LABEL: .func{{.*}}shfl_sync_rrr
|
|
define i32 @shfl_sync_rrr(i32 %mask, i32 %a, i32 %b, i32 %c) {
|
|
; CHECK: ld.param.b32 [[MASK:%r[0-9]+]]
|
|
; CHECK: ld.param.b32 [[A:%r[0-9]+]]
|
|
; CHECK: ld.param.b32 [[B:%r[0-9]+]]
|
|
; CHECK: ld.param.b32 [[C:%r[0-9]+]]
|
|
; CHECK: shfl.sync.down.b32 [[OUT:%r[0-9]+]], [[A]], [[B]], [[C]], [[MASK]];
|
|
; CHECK: st.param.{{.}}32 {{.*}}, [[OUT]]
|
|
%val = call i32 @llvm.nvvm.shfl.sync.down.i32(i32 %mask, i32 %a, i32 %b, i32 %c)
|
|
ret i32 %val
|
|
}
|
|
|
|
; CHECK-LABEL: .func{{.*}}shfl_sync_irr
|
|
define i32 @shfl_sync_irr(i32 %a, i32 %b, i32 %c) {
|
|
; CHECK: ld.param.b32 [[A:%r[0-9]+]]
|
|
; CHECK: ld.param.b32 [[B:%r[0-9]+]]
|
|
; CHECK: ld.param.b32 [[C:%r[0-9]+]]
|
|
; CHECK: shfl.sync.down.b32 [[OUT:%r[0-9]+]], [[A]], [[B]], [[C]], 1;
|
|
; CHECK: st.param.{{.}}32 {{.*}}, [[OUT]]
|
|
%val = call i32 @llvm.nvvm.shfl.sync.down.i32(i32 1, i32 %a, i32 %b, i32 %c)
|
|
ret i32 %val
|
|
}
|
|
|
|
; CHECK-LABEL: .func{{.*}}shfl_sync_rri
|
|
define i32 @shfl_sync_rri(i32 %mask, i32 %a, i32 %b) {
|
|
; CHECK: ld.param.b32 [[MASK:%r[0-9]+]]
|
|
; CHECK: ld.param.b32 [[A:%r[0-9]+]]
|
|
; CHECK: ld.param.b32 [[B:%r[0-9]+]]
|
|
; CHECK: shfl.sync.down.b32 [[OUT:%r[0-9]+]], [[A]], [[B]], 1, [[MASK]];
|
|
; CHECK: st.param.{{.}}32 {{.*}}, [[OUT]]
|
|
%val = call i32 @llvm.nvvm.shfl.sync.down.i32(i32 %mask, i32 %a, i32 %b, i32 1)
|
|
ret i32 %val
|
|
}
|
|
|
|
; CHECK-LABEL: .func{{.*}}shfl_sync_iri
|
|
define i32 @shfl_sync_iri(i32 %a, i32 %b) {
|
|
; CHECK: ld.param.b32 [[A:%r[0-9]+]]
|
|
; CHECK: ld.param.b32 [[B:%r[0-9]+]]
|
|
; CHECK: shfl.sync.down.b32 [[OUT:%r[0-9]+]], [[A]], [[B]], 2, 1;
|
|
; CHECK: st.param.{{.}}32 {{.*}}, [[OUT]]
|
|
%val = call i32 @llvm.nvvm.shfl.sync.down.i32(i32 1, i32 %a, i32 %b, i32 2)
|
|
ret i32 %val
|
|
}
|
|
|
|
; CHECK-LABEL: .func{{.*}}shfl_sync_rir
|
|
define i32 @shfl_sync_rir(i32 %mask, i32 %a, i32 %c) {
|
|
; CHECK: ld.param.b32 [[MASK:%r[0-9]+]]
|
|
; CHECK: ld.param.b32 [[A:%r[0-9]+]]
|
|
; CHECK: ld.param.b32 [[C:%r[0-9]+]]
|
|
; CHECK: shfl.sync.down.b32 [[OUT:%r[0-9]+]], [[A]], 1, [[C]], [[MASK]];
|
|
; CHECK: st.param.{{.}}32 {{.*}}, [[OUT]]
|
|
%val = call i32 @llvm.nvvm.shfl.sync.down.i32(i32 %mask, i32 %a, i32 1, i32 %c)
|
|
ret i32 %val
|
|
}
|
|
|
|
; CHECK-LABEL: .func{{.*}}shfl_sync_iir
|
|
define i32 @shfl_sync_iir(i32 %a, i32 %c) {
|
|
; CHECK: ld.param.b32 [[A:%r[0-9]+]]
|
|
; CHECK: ld.param.b32 [[C:%r[0-9]+]]
|
|
; CHECK: shfl.sync.down.b32 [[OUT:%r[0-9]+]], [[A]], 2, [[C]], 1;
|
|
; CHECK: st.param.{{.}}32 {{.*}}, [[OUT]]
|
|
%val = call i32 @llvm.nvvm.shfl.sync.down.i32(i32 1, i32 %a, i32 2, i32 %c)
|
|
ret i32 %val
|
|
}
|
|
|
|
; CHECK-LABEL: .func{{.*}}shfl_sync_rii
|
|
define i32 @shfl_sync_rii(i32 %mask, i32 %a) {
|
|
; CHECK: ld.param.b32 [[MASK:%r[0-9]+]]
|
|
; CHECK: ld.param.b32 [[A:%r[0-9]+]]
|
|
; CHECK: shfl.sync.down.b32 [[OUT:%r[0-9]+]], [[A]], 1, 2, [[MASK]];
|
|
; CHECK: st.param.{{.}}32 {{.*}}, [[OUT]]
|
|
%val = call i32 @llvm.nvvm.shfl.sync.down.i32(i32 %mask, i32 %a, i32 1, i32 2)
|
|
ret i32 %val
|
|
}
|
|
|
|
; CHECK-LABEL: .func{{.*}}shfl_sync_iii
|
|
define i32 @shfl_sync_iii(i32 %a, i32 %b) {
|
|
; CHECK: ld.param.b32 [[A:%r[0-9]+]]
|
|
; CHECK: shfl.sync.down.b32 [[OUT:%r[0-9]+]], [[A]], 2, 3, 1;
|
|
; CHECK: st.param.{{.}}32 {{.*}}, [[OUT]]
|
|
%val = call i32 @llvm.nvvm.shfl.sync.down.i32(i32 1, i32 %a, i32 2, i32 3)
|
|
ret i32 %val
|
|
}
|