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
134 lines
5.2 KiB
LLVM
134 lines
5.2 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
|
|
; RUN: llc < %s -mtriple=nvptx64 -mcpu=sm_100a -mattr=+ptx86 | FileCheck %s
|
|
; RUN: llc < %s -mtriple=nvptx64 -mcpu=sm_100f -mattr=+ptx88 | FileCheck %s
|
|
; RUN: %if ptxas-sm_100a && ptxas-isa-8.6 %{ llc < %s -mtriple=nvptx64 -mcpu=sm_100a -mattr=+ptx86 | %ptxas-verify -arch=sm_100a %}
|
|
; RUN: %if ptxas-sm_100f && ptxas-isa-8.8 %{ llc < %s -mtriple=nvptx64 -mcpu=sm_100f -mattr=+ptx88 | %ptxas-verify -arch=sm_100f %}
|
|
|
|
declare float @llvm.nvvm.redux.sync.fmin(float, i32)
|
|
define float @redux_sync_fmin(float %src, i32 %mask) {
|
|
; CHECK-LABEL: redux_sync_fmin(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<4>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [redux_sync_fmin_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [redux_sync_fmin_param_1];
|
|
; CHECK-NEXT: redux.sync.min.f32 %r3, %r1, %r2;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r3;
|
|
; CHECK-NEXT: ret;
|
|
%val = call float @llvm.nvvm.redux.sync.fmin(float %src, i32 %mask)
|
|
ret float %val
|
|
}
|
|
|
|
declare float @llvm.nvvm.redux.sync.fmin.abs(float, i32)
|
|
define float @redux_sync_fmin_abs(float %src, i32 %mask) {
|
|
; CHECK-LABEL: redux_sync_fmin_abs(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<4>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [redux_sync_fmin_abs_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [redux_sync_fmin_abs_param_1];
|
|
; CHECK-NEXT: redux.sync.min.abs.f32 %r3, %r1, %r2;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r3;
|
|
; CHECK-NEXT: ret;
|
|
%val = call float @llvm.nvvm.redux.sync.fmin.abs(float %src, i32 %mask)
|
|
ret float %val
|
|
}
|
|
|
|
declare float @llvm.nvvm.redux.sync.fmin.NaN(float, i32)
|
|
define float @redux_sync_fmin_NaN(float %src, i32 %mask) {
|
|
; CHECK-LABEL: redux_sync_fmin_NaN(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<4>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [redux_sync_fmin_NaN_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [redux_sync_fmin_NaN_param_1];
|
|
; CHECK-NEXT: redux.sync.min.NaN.f32 %r3, %r1, %r2;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r3;
|
|
; CHECK-NEXT: ret;
|
|
%val = call float @llvm.nvvm.redux.sync.fmin.NaN(float %src, i32 %mask)
|
|
ret float %val
|
|
}
|
|
|
|
declare float @llvm.nvvm.redux.sync.fmin.abs.NaN(float, i32)
|
|
define float @redux_sync_fmin_abs_NaN(float %src, i32 %mask) {
|
|
; CHECK-LABEL: redux_sync_fmin_abs_NaN(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<4>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [redux_sync_fmin_abs_NaN_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [redux_sync_fmin_abs_NaN_param_1];
|
|
; CHECK-NEXT: redux.sync.min.abs.NaN.f32 %r3, %r1, %r2;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r3;
|
|
; CHECK-NEXT: ret;
|
|
%val = call float @llvm.nvvm.redux.sync.fmin.abs.NaN(float %src, i32 %mask)
|
|
ret float %val
|
|
}
|
|
|
|
declare float @llvm.nvvm.redux.sync.fmax(float, i32)
|
|
define float @redux_sync_fmax(float %src, i32 %mask) {
|
|
; CHECK-LABEL: redux_sync_fmax(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<4>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [redux_sync_fmax_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [redux_sync_fmax_param_1];
|
|
; CHECK-NEXT: redux.sync.max.f32 %r3, %r1, %r2;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r3;
|
|
; CHECK-NEXT: ret;
|
|
%val = call float @llvm.nvvm.redux.sync.fmax(float %src, i32 %mask)
|
|
ret float %val
|
|
}
|
|
|
|
declare float @llvm.nvvm.redux.sync.fmax.abs(float, i32)
|
|
define float @redux_sync_fmax_abs(float %src, i32 %mask) {
|
|
; CHECK-LABEL: redux_sync_fmax_abs(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<4>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [redux_sync_fmax_abs_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [redux_sync_fmax_abs_param_1];
|
|
; CHECK-NEXT: redux.sync.max.abs.f32 %r3, %r1, %r2;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r3;
|
|
; CHECK-NEXT: ret;
|
|
%val = call float @llvm.nvvm.redux.sync.fmax.abs(float %src, i32 %mask)
|
|
ret float %val
|
|
}
|
|
|
|
declare float @llvm.nvvm.redux.sync.fmax.NaN(float, i32)
|
|
define float @redux_sync_fmax_NaN(float %src, i32 %mask) {
|
|
; CHECK-LABEL: redux_sync_fmax_NaN(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<4>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [redux_sync_fmax_NaN_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [redux_sync_fmax_NaN_param_1];
|
|
; CHECK-NEXT: redux.sync.max.NaN.f32 %r3, %r1, %r2;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r3;
|
|
; CHECK-NEXT: ret;
|
|
%val = call float @llvm.nvvm.redux.sync.fmax.NaN(float %src, i32 %mask)
|
|
ret float %val
|
|
}
|
|
|
|
declare float @llvm.nvvm.redux.sync.fmax.abs.NaN(float, i32)
|
|
define float @redux_sync_fmax_abs_NaN(float %src, i32 %mask) {
|
|
; CHECK-LABEL: redux_sync_fmax_abs_NaN(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<4>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [redux_sync_fmax_abs_NaN_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [redux_sync_fmax_abs_NaN_param_1];
|
|
; CHECK-NEXT: redux.sync.max.abs.NaN.f32 %r3, %r1, %r2;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r3;
|
|
; CHECK-NEXT: ret;
|
|
%val = call float @llvm.nvvm.redux.sync.fmax.abs.NaN(float %src, i32 %mask)
|
|
ret float %val
|
|
}
|