search

MARIO-Math-Reasoning / vllm

Apache License 2.0
3 stars 1 forks source link

Building CUDA object Error #2

Open CN-Wenbo opened 1 month ago

CN-Wenbo commented 1 month ago

Thanks for sharing the vllm project. The following problem have caused me building failed while executing "pip install ". IMG_8469

CUDA_VERSION 12.2 CMAKE_VERSION 3.22.1 GPU Tesla P100

lovecambi commented 1 month ago

You should make your GPU match its corresponding compute capability. According to Nvidia document -> CUDA-Enabled Datacenter Products -> NVIDIA Data Center Products, your Tesla P100 matches 6.0. But the error message shows your arch is less than 5.3.