Avoid scalar indexing with GPU arrays #40
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Codecov Report
@@ Coverage Diff @@
## master #40 +/- ##
==========================================
+ Coverage 97.15% 97.17% +0.02%
==========================================
Files 17 18 +1
Lines 983 1026 +43
==========================================
+ Hits 955 997 +42
- Misses 28 29 +1
Continue to review full report at Codecov.
|
For now scalar indexing can be avoided when dimension permutations are disabled...
jipolanco
force-pushed
the
gpu-indexing
branch
from
dc356a9 to
e6f69a4
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
For GPU arrays, transpositions and other operations are now performed completely on the GPU (as far as I can tell...), avoiding slow scalar indexing.
Well, for now this has just been tested with the reference implementation of GPUArrays.jl (
JLArray), which is implemented on CPUs.It would be nice to test things with
CuArrays. For that, one just needs to addCuArrayto the list of array types tested intest/array_types.jl. @corentin-dev let me know if you can try that out.For now I have no idea how the transposition of GPU arrays actually performs, and it would be nice to have some benchmarks. There are still some things that can be improved. In particular, when using dimension permutations (enabled by default in PencilFFTs), there are some additional allocations that should be taken care of.
This PR closes #21 (but can be reopened if stuff is missing).