pytorch

ae6dd20b - [cuDNN V8 API] (reopen 2) Allow the number of kernels profiled under torch.backends.cudnn.benchmark = True to be limitedCudnnv8 benchmark limit (#78299)

[cuDNN V8 API] (reopen 2) Allow the number of kernels profiled under torch.backends.cudnn.benchmark = True to be limitedCudnnv8 benchmark limit (#78299) Reopen of #77002 to address comments by @malfet CC @ngimel @ptrblck Pull Request resolved: https://github.com/pytorch/pytorch/pull/78299 Approved by: https://github.com/ngimel