Where to find detailed information about the algorithm type when using cublasGemmEx

The cublas doc provide a brief meaning of cublasGemmAlgo_t, is there some detailed information about these algorithm value or is it not public yet?

It’s not public, other than what is in the cublas doc you linked. There isn’t another document to refer to, other than the cublas doc, to get information about cublas. If you have specific questions, you can ask them here, and/or you can request changes to the CUDA documentation by filing a bug.