NVIDIA/TensorRT-LLM

why so many kernels use cubin? fully open-source?

Opened this issue · 2 comments

As the new show:
[03/22] TensorRT-LLM is now fully open-source, with developments moved to GitHub!

BUT,why so many kernels use cubin? like this: cubin
Thank you!

When I want to use TRT-LLM in THOR platform, it is so hard.

When I want to use TRT-LLM in THOR platform, it is so hard.当我想在 THOR 平台上使用 TRT-LLM 时,这太难了。

I have same question...