Skip to content
Cuda Test

LLM GEMM benchmarks #280

Sign in to view logs

LLM GEMM benchmarks

LLM GEMM benchmarks #280

Workflow file for this run

.github/workflows/cuda_test.yml at 1f4e5cf
name: "Cuda Test"
on:
push:
branches: [ "sycl-develop" ]
pull_request:
branches: [ "sycl-develop" ]
workflow_dispatch:
permissions: {}
jobs:
run-tests:
name: Run cuda tests
runs-on: cp-nvidia-gpu
timeout-minutes: 120
steps:
- name: Checkout repository
uses: actions/checkout@a5ac7e51b41094c92402da3b24376905380afc29 # v4.1.6
- name: Build
shell: bash
run: |
nvidia-smi
export CUDACXX=/usr/local/cuda/bin/nvcc
cmake -G Ninja -DCUTLASS_NVCC_ARCHS="90a"
cmake --build .
- name: Unit test
shell: bash
run: |
cmake --build . --target test_unit -j