Open fnscar opened 5 years ago
Hope to hear more on this! The potential of Tensor cores on the RTX 2XXX series seems very intriguing!
In case someone would want to integrate this. It seems relatively straight forward on CuDNN 7.6.5 and above.
Relevant documentation overview: https://docs.nvidia.com/deeplearning/sdk/cudnn-developer-guide/index.html#tensor-ops-conv-functions
お金に余裕ができてTensorコアを搭載したグラボを購入出来たら検証予定です。