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
116 lines
4.4 KiB
LLVM
116 lines
4.4 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
|
|
; RUN: llc < %s -mtriple=nvptx64 -mcpu=sm_20 -mattr=+ptx42 | FileCheck %s
|
|
; RUN: %if ptxas-sm_20 %{ llc < %s -mtriple=nvptx64 -mcpu=sm_20 | %ptxas-verify -arch=sm_20 %}
|
|
|
|
define float @add_sat_f32(float %a, float %b) {
|
|
; CHECK-LABEL: add_sat_f32(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<11>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [add_sat_f32_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [add_sat_f32_param_1];
|
|
; CHECK-NEXT: add.rn.sat.f32 %r3, %r1, %r2;
|
|
; CHECK-NEXT: add.rn.sat.ftz.f32 %r4, %r1, %r3;
|
|
; CHECK-NEXT: add.rz.sat.f32 %r5, %r1, %r4;
|
|
; CHECK-NEXT: add.rz.sat.ftz.f32 %r6, %r1, %r5;
|
|
; CHECK-NEXT: add.rm.sat.f32 %r7, %r1, %r6;
|
|
; CHECK-NEXT: add.rm.sat.ftz.f32 %r8, %r1, %r7;
|
|
; CHECK-NEXT: add.rp.sat.f32 %r9, %r1, %r8;
|
|
; CHECK-NEXT: add.rp.sat.ftz.f32 %r10, %r1, %r9;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r10;
|
|
; CHECK-NEXT: ret;
|
|
%r1 = call float @llvm.nvvm.add.rn.sat.f(float %a, float %b)
|
|
%r2 = call float @llvm.nvvm.add.rn.ftz.sat.f(float %a, float %r1)
|
|
|
|
%r3 = call float @llvm.nvvm.add.rz.sat.f(float %a, float %r2)
|
|
%r4 = call float @llvm.nvvm.add.rz.ftz.sat.f(float %a, float %r3)
|
|
|
|
%r5 = call float @llvm.nvvm.add.rm.sat.f(float %a, float %r4)
|
|
%r6 = call float @llvm.nvvm.add.rm.ftz.sat.f(float %a, float %r5)
|
|
|
|
%r7 = call float @llvm.nvvm.add.rp.sat.f(float %a, float %r6)
|
|
%r8 = call float @llvm.nvvm.add.rp.ftz.sat.f(float %a, float %r7)
|
|
|
|
ret float %r8
|
|
}
|
|
|
|
define float @sub_sat_f32(float %a, float %b) {
|
|
; CHECK-LABEL: sub_sat_f32(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<11>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [sub_sat_f32_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [sub_sat_f32_param_1];
|
|
; CHECK-NEXT: sub.rn.sat.f32 %r3, %r1, %r2;
|
|
; CHECK-NEXT: sub.rn.sat.ftz.f32 %r4, %r1, %r3;
|
|
; CHECK-NEXT: sub.rz.sat.f32 %r5, %r1, %r4;
|
|
; CHECK-NEXT: sub.rz.sat.ftz.f32 %r6, %r1, %r5;
|
|
; CHECK-NEXT: sub.rm.sat.f32 %r7, %r1, %r6;
|
|
; CHECK-NEXT: sub.rm.sat.ftz.f32 %r8, %r1, %r7;
|
|
; CHECK-NEXT: sub.rp.sat.f32 %r9, %r1, %r8;
|
|
; CHECK-NEXT: sub.rp.sat.ftz.f32 %r10, %r1, %r9;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r10;
|
|
; CHECK-NEXT: ret;
|
|
%f0 = fneg float %b
|
|
%r1 = call float @llvm.nvvm.add.rn.sat.f(float %a, float %f0)
|
|
|
|
%f1 = fneg float %r1
|
|
%r2 = call float @llvm.nvvm.add.rn.ftz.sat.f(float %a, float %f1)
|
|
|
|
%f2 = fneg float %r2
|
|
%r3 = call float @llvm.nvvm.add.rz.sat.f(float %a, float %f2)
|
|
|
|
%f3 = fneg float %r3
|
|
%r4 = call float @llvm.nvvm.add.rz.ftz.sat.f(float %a, float %f3)
|
|
|
|
%f4 = fneg float %r4
|
|
%r5 = call float @llvm.nvvm.add.rm.sat.f(float %a, float %f4)
|
|
|
|
%f5 = fneg float %r5
|
|
%r6 = call float @llvm.nvvm.add.rm.ftz.sat.f(float %a, float %f5)
|
|
|
|
%f6 = fneg float %r6
|
|
%r7 = call float @llvm.nvvm.add.rp.sat.f(float %a, float %f6)
|
|
|
|
%f7 = fneg float %r7
|
|
%r8 = call float @llvm.nvvm.add.rp.ftz.sat.f(float %a, float %f7)
|
|
|
|
ret float %r8
|
|
}
|
|
|
|
define float @fma_sat_f32(float %a, float %b, float %c) {
|
|
; CHECK-LABEL: fma_sat_f32(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<12>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [fma_sat_f32_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [fma_sat_f32_param_1];
|
|
; CHECK-NEXT: ld.param.b32 %r3, [fma_sat_f32_param_2];
|
|
; CHECK-NEXT: fma.rn.sat.f32 %r4, %r1, %r2, %r3;
|
|
; CHECK-NEXT: fma.rn.ftz.sat.f32 %r5, %r1, %r2, %r4;
|
|
; CHECK-NEXT: fma.rz.sat.f32 %r6, %r1, %r2, %r5;
|
|
; CHECK-NEXT: fma.rz.ftz.sat.f32 %r7, %r1, %r2, %r6;
|
|
; CHECK-NEXT: fma.rm.sat.f32 %r8, %r1, %r2, %r7;
|
|
; CHECK-NEXT: fma.rm.ftz.sat.f32 %r9, %r1, %r2, %r8;
|
|
; CHECK-NEXT: fma.rp.sat.f32 %r10, %r1, %r2, %r9;
|
|
; CHECK-NEXT: fma.rp.ftz.sat.f32 %r11, %r1, %r2, %r10;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r11;
|
|
; CHECK-NEXT: ret;
|
|
%r1 = call float @llvm.nvvm.fma.rn.sat.f(float %a, float %b, float %c)
|
|
%r2 = call float @llvm.nvvm.fma.rn.ftz.sat.f(float %a, float %b, float %r1)
|
|
|
|
%r3 = call float @llvm.nvvm.fma.rz.sat.f(float %a, float %b, float %r2)
|
|
%r4 = call float @llvm.nvvm.fma.rz.ftz.sat.f(float %a, float %b, float %r3)
|
|
|
|
%r5 = call float @llvm.nvvm.fma.rm.sat.f(float %a, float %b, float %r4)
|
|
%r6 = call float @llvm.nvvm.fma.rm.ftz.sat.f(float %a, float %b, float %r5)
|
|
|
|
%r7 = call float @llvm.nvvm.fma.rp.sat.f(float %a, float %b, float %r6)
|
|
%r8 = call float @llvm.nvvm.fma.rp.ftz.sat.f(float %a, float %b, float %r7)
|
|
|
|
ret float %r8
|
|
}
|