@samskalicky commented on this pull request.
> @@ -42,15 +43,18 @@ set(USE_CUDA ON CACHE BOOL "Build with CUDA support") set(USE_CUDNN ON CACHE BOOL "Build with cudnn support, if found") # Target NVIDIA GPU achitecture. -# Valid options are "Auto" for autodetection, "All" for all available -# architectures or a list of architectures by compute capability number, such as -# "7.0" or "7.0;7.5" as well as name, such as "Volta" or "Volta;Turing". +# Valid options are: +# - "Auto" for autodetection, will try and discover which GPU architecture to use by +# looking at the available GPUs on the machine that you're building on +# - "All" for all available GPU architectures supported by the version of CUDA installed +# - "specific GPU architectures" by giving the compute capability number such as +# "7.0" or "7.0;7.5" (ie. sm_70 or sm_75) or you can specify the name like: +# "Volta" or "Volta;Turing". @leezu I tried to merge my version which what was there before. Hows this? -- You are receiving this because you were mentioned. Reply to this email directly or view it on GitHub: https://github.com/apache/incubator-mxnet/pull/18904#pullrequestreview-474812452
