kehuanfeng opened a new pull request #8265: URL: https://github.com/apache/tvm/pull/8265
The shape parameters for filter are not assigned properly when inferencing with int8 and cudnn. cudnnStatus_t cudnnSetFilter4dDescriptor( cudnnFilterDescriptor_t filterDesc, cudnnDataType_t dataType, cudnnTensorFormat_t format, int k, int c, int h, int w) Above is the definition from cudnn dev guide (k is the output channel which should be from the w->shape[3] and c is the input channel which shoudl be from w->shape[2]) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
