Compatibility with GPU capability 70 and below #22
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes #12 #13 #17
Adds support for Volta tensor cores via a new
CutlassMLP
configuration.FullyFusedMLP
automatically falls back toCutlassMLP
when a compute capability of 70 and below is requested. This means that there'll be a performance regression when targeting multiple compute capabilities at the same time (the lowest one will determine the compatibility setting).Also enables the use of full-precision CUDA-core matrix multiplication for compute capabilities below 70, i.e. 61 and lower.