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
132 lines
5.2 KiB
LLVM
132 lines
5.2 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_90 -mattr=+ptx81 | FileCheck %s
|
|
; RUN: %if ptxas-isa-8.1 %{ llc < %s -mtriple=nvptx64 -mcpu=sm_90 -mattr=+ptx81 | %ptxas-verify -arch=sm_90 %}
|
|
|
|
define half @fma_oob_f16(half %a, half %b, half %c) {
|
|
; CHECK-LABEL: fma_oob_f16(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b16 %rs<5>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b16 %rs1, [fma_oob_f16_param_0];
|
|
; CHECK-NEXT: ld.param.b16 %rs2, [fma_oob_f16_param_1];
|
|
; CHECK-NEXT: ld.param.b16 %rs3, [fma_oob_f16_param_2];
|
|
; CHECK-NEXT: fma.rn.oob.f16 %rs4, %rs1, %rs2, %rs3;
|
|
; CHECK-NEXT: st.param.b16 [func_retval0], %rs4;
|
|
; CHECK-NEXT: ret;
|
|
%1 = call half @llvm.nvvm.fma.rn.oob.f16(half %a, half %b, half %c)
|
|
ret half %1
|
|
}
|
|
|
|
define half @fma_oob_relu_f16(half %a, half %b, half %c) {
|
|
; CHECK-LABEL: fma_oob_relu_f16(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b16 %rs<5>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b16 %rs1, [fma_oob_relu_f16_param_0];
|
|
; CHECK-NEXT: ld.param.b16 %rs2, [fma_oob_relu_f16_param_1];
|
|
; CHECK-NEXT: ld.param.b16 %rs3, [fma_oob_relu_f16_param_2];
|
|
; CHECK-NEXT: fma.rn.oob.relu.f16 %rs4, %rs1, %rs2, %rs3;
|
|
; CHECK-NEXT: st.param.b16 [func_retval0], %rs4;
|
|
; CHECK-NEXT: ret;
|
|
%1 = call half @llvm.nvvm.fma.rn.oob.relu.f16(half %a, half %b, half %c)
|
|
ret half %1
|
|
}
|
|
|
|
define <2 x half> @fma_oob_f16x2(<2 x half> %a, <2 x half> %b, <2 x half> %c) {
|
|
; CHECK-LABEL: fma_oob_f16x2(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<5>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [fma_oob_f16x2_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [fma_oob_f16x2_param_1];
|
|
; CHECK-NEXT: ld.param.b32 %r3, [fma_oob_f16x2_param_2];
|
|
; CHECK-NEXT: fma.rn.oob.f16x2 %r4, %r1, %r2, %r3;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r4;
|
|
; CHECK-NEXT: ret;
|
|
%1 = call <2 x half> @llvm.nvvm.fma.rn.oob.v2f16( <2 x half> %a, <2 x half> %b, <2 x half> %c)
|
|
ret <2 x half> %1
|
|
}
|
|
|
|
define <2 x half> @fma_oob_relu_f16x2(<2 x half> %a, <2 x half> %b, <2 x half> %c) {
|
|
; CHECK-LABEL: fma_oob_relu_f16x2(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<5>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [fma_oob_relu_f16x2_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [fma_oob_relu_f16x2_param_1];
|
|
; CHECK-NEXT: ld.param.b32 %r3, [fma_oob_relu_f16x2_param_2];
|
|
; CHECK-NEXT: fma.rn.oob.relu.f16x2 %r4, %r1, %r2, %r3;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r4;
|
|
; CHECK-NEXT: ret;
|
|
%1 = call <2 x half> @llvm.nvvm.fma.rn.oob.relu.v2f16( <2 x half> %a, <2 x half> %b, <2 x half> %c)
|
|
ret <2 x half> %1
|
|
}
|
|
|
|
define bfloat @fma_oob_bf16(bfloat %a, bfloat %b, bfloat %c) {
|
|
; CHECK-LABEL: fma_oob_bf16(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b16 %rs<5>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b16 %rs1, [fma_oob_bf16_param_0];
|
|
; CHECK-NEXT: ld.param.b16 %rs2, [fma_oob_bf16_param_1];
|
|
; CHECK-NEXT: ld.param.b16 %rs3, [fma_oob_bf16_param_2];
|
|
; CHECK-NEXT: fma.rn.oob.bf16 %rs4, %rs1, %rs2, %rs3;
|
|
; CHECK-NEXT: st.param.b16 [func_retval0], %rs4;
|
|
; CHECK-NEXT: ret;
|
|
%1 = call bfloat @llvm.nvvm.fma.rn.oob.bf16(bfloat %a, bfloat %b, bfloat %c)
|
|
ret bfloat %1
|
|
}
|
|
|
|
define bfloat @fma_oob_relu_bf16(bfloat %a, bfloat %b, bfloat %c) {
|
|
; CHECK-LABEL: fma_oob_relu_bf16(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b16 %rs<5>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b16 %rs1, [fma_oob_relu_bf16_param_0];
|
|
; CHECK-NEXT: ld.param.b16 %rs2, [fma_oob_relu_bf16_param_1];
|
|
; CHECK-NEXT: ld.param.b16 %rs3, [fma_oob_relu_bf16_param_2];
|
|
; CHECK-NEXT: fma.rn.oob.relu.bf16 %rs4, %rs1, %rs2, %rs3;
|
|
; CHECK-NEXT: st.param.b16 [func_retval0], %rs4;
|
|
; CHECK-NEXT: ret;
|
|
%1 = call bfloat @llvm.nvvm.fma.rn.oob.relu.bf16(bfloat %a, bfloat %b, bfloat %c)
|
|
ret bfloat %1
|
|
}
|
|
|
|
define <2 x bfloat> @fma_oob_bf16x2(<2 x bfloat> %a, <2 x bfloat> %b, <2 x bfloat> %c) {
|
|
; CHECK-LABEL: fma_oob_bf16x2(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<5>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [fma_oob_bf16x2_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [fma_oob_bf16x2_param_1];
|
|
; CHECK-NEXT: ld.param.b32 %r3, [fma_oob_bf16x2_param_2];
|
|
; CHECK-NEXT: fma.rn.oob.bf16x2 %r4, %r1, %r2, %r3;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r4;
|
|
; CHECK-NEXT: ret;
|
|
%1 = call <2 x bfloat> @llvm.nvvm.fma.rn.oob.v2bf16( <2 x bfloat> %a, <2 x bfloat> %b, <2 x bfloat> %c)
|
|
ret <2 x bfloat> %1
|
|
}
|
|
|
|
define <2 x bfloat> @fma_oob_relu_bf16x2(<2 x bfloat> %a, <2 x bfloat> %b, <2 x bfloat> %c) {
|
|
; CHECK-LABEL: fma_oob_relu_bf16x2(
|
|
; CHECK: {
|
|
; CHECK-NEXT: .reg .b32 %r<5>;
|
|
; CHECK-EMPTY:
|
|
; CHECK-NEXT: // %bb.0:
|
|
; CHECK-NEXT: ld.param.b32 %r1, [fma_oob_relu_bf16x2_param_0];
|
|
; CHECK-NEXT: ld.param.b32 %r2, [fma_oob_relu_bf16x2_param_1];
|
|
; CHECK-NEXT: ld.param.b32 %r3, [fma_oob_relu_bf16x2_param_2];
|
|
; CHECK-NEXT: fma.rn.oob.relu.bf16x2 %r4, %r1, %r2, %r3;
|
|
; CHECK-NEXT: st.param.b32 [func_retval0], %r4;
|
|
; CHECK-NEXT: ret;
|
|
%1 = call <2 x bfloat> @llvm.nvvm.fma.rn.oob.relu.v2bf16( <2 x bfloat> %a, <2 x bfloat> %b, <2 x bfloat> %c)
|
|
ret <2 x bfloat> %1
|
|
}
|