[BUG] does not support nvidia A30

microsoft / nnfusion

A flexible and efficient deep neural network (DNN) compiler that generates high-performance executable from a DNN model description.

MIT License

948 stars 158 forks source link

[BUG] does not support nvidia A30 #450

Open zhaohb opened 2 years ago

zhaohb commented 2 years ago

I converted a model on the A30 card, but was stuck in cuda_init when main_test was executed. It's stuck on:

...
 CUDNN_SAFE_CALL(cudnnCreate(&cudnn_handle_0))
...

how to fix it?

xysmlx commented 2 years ago

Hi, @zhaohb,

Thank you for your attention. You may need to add the compilation information (i.e., -gencode arch=compute_80,code=sm_80) to the compilation flags to support the A30 GPU, and re-build the NNFusion.

puddingfjz commented 1 year ago

Hi, @xysmlx,

If I want to try Roller on A100, which part of the code should I modify?