search

flashinfer-ai / flashinfer

FlashInfer: Kernel Library for LLM Serving
https://flashinfer.ai
Apache License 2.0
768 stars 64 forks source link

misc: add some notes in `cmake.config` #293

Closed yzh119 closed 4 weeks ago

yzh119 commented 4 weeks ago

We found that setting CUDA architecture to native may result in slow compilation, especially if there are GPU programs running. This PR adds some notes to the config.cmake file to encourage user to set the architecture to a specific value.