Extend GPU support to Metal, ROCm, and oneAPI backends #405
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
…ckageExtensionCompat dependency
…ing raw acquisition data
… to fix Metal issue
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #405 +/- ##
==========================================
- Coverage 89.47% 89.32% -0.16%
==========================================
Files 43 44 +1
Lines 2728 2736 +8
==========================================
+ Hits 2441 2444 +3
- Misses 287 292 +5
Flags with carried forward coverage won't be shown. Click here to find out more.
|
cncastillo
requested changes
Jun 12, 2024
There was a problem hiding this comment.
Perfect! add print_devices in docs/src/reference/3-koma-core.md so the documentation doesn't fail.
Maybe something like this (I renamed the subsection)
## GPU helper functions
```@docs
print_devices
gpu
cpu
f32
f64```Should we add a warning to KomaMetalExt.__init__ saying that performance could be affected by JuliaGPU/Metal.jl#348?
Merged
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Outdated suggestions cannot be applied.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.
This pull request adds support for GPU backends Metal, ROCm, and oneAPI with the corresponding Julia packages (Metal.jl, AMDGPU.jl, oneAPI.jl). CUDA is still supported but is no longer an explicit dependency, so users will now need to add 'using CUDA' to their code or they will see a message informing them that no GPU backends are loaded. I've run the KomaMRICore tests with CUDA and Metal and ensured that they pass, but for AMD and oneAPI we will need to wait until
#147 to confirm that this works as expected.
I haven't made any documentation updates yet, so let me know if that needs to be done before merging. The last thing to note is with the ArbitraryMotion struct, there was a custom adapt_storage function for adapting to a CUDA device that I wasn't sure whether it was necessary, since I think Adapt.jl and Functors.jl should be able to adapt the fields correctly and the ArbitraryMotion GPU test still passed with my changes in place. Nevertheless, it's possible I will need to add that function back, and define adapt_storage functions for converting ArbitraryMotion to the other backends.