News for hackers. Stuff that matters.

A comment on Pascal GPUs on All Fronts Push Nvidia to New Highs. Parent comment

Re:Pascal GPUs on All Fronts Push Nvidia to New Highs by gcr on Monday November 14, 2016 @09:51AM (#12950101)
HPC? NVidia is completely destroying AMD on the deep learning front. The most popular deep learning frameworks use CUDA and do not have stable OpenCL support out of the box.
[ Parent | on HN ]
Re:Pascal GPUs on All Fronts Push Nvidia to New Highs by gcp on Monday November 14, 2016 @10:51AM (#12950542)
The most popular deep learning frameworks use CUDA and do not have stable OpenCL support out of the box.

It's worse than that. The popular frameworks use cuDNN, which is a hand (assembler) optimized library for common operations. AMD does not have any equivalent. Even if you have OpenCL support you'll get nowhere the performance.

[ Parent | on HN ]
Re:Pascal GPUs on All Fronts Push Nvidia to New Highs by gcr on Monday November 14, 2016 @03:03PM (#12952776)
Most libraries have fallbacks for cuDNN-less operation. Torch, for example, requires you to import a CUDNN library if you want to use it. Caffe requires you to uncomment a line in a makefile if you want to use it.
[ Parent | on HN ]
Re:Pascal GPUs on All Fronts Push Nvidia to New Highs by gcp on Tuesday November 15, 2016 @04:20AM (#12956684)
As already said, you'll get nowhere the performance.
[ Parent | on HN ]
Re:Pascal GPUs on All Fronts Push Nvidia to New Highs by bryanlarsen on Monday November 14, 2016 @10:23AM (#12950319)
And today AMD announced support for CUDA and hinted that they were deprecating OpenCL (by releasing OpenCL 1.2+ rather than concentrating on 2.0).
[ Parent | on HN ]
Re:Pascal GPUs on All Fronts Push Nvidia to New Highs by gcp on Monday November 14, 2016 @10:47AM (#12950489)
What? AMD's drivers have had OpenCL 2.0 support for a long time. At least on Windows.
[ Parent | on HN ]