adds mappings for f16 variants of halide float math #8029
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
steven-johnson
approved these changes
Jan 10, 2024
Im happy to write one... though I didn't see any coverage of the existing f32 codegen ( correctness/float16_t.cpp has coverage of those intrinsics, but only under cpu scheduling, and a simple test of metal roundtripping ) |
Doesn't need to be done as part of this PR, but IMHO we clearly need to extend this test to be executed under GPU since we expect those features to work. If you don't have the bandwidth to do so, let's get someone else to do it.. |
|
Update: I'm gonna land this and take it upon myself to add the right tests |
steven-johnson
added a commit
that referenced
this pull request
Jan 16, 2024
Tweak correctness_float16_t so that it uses one of the transcendal functions (sqrt) that were missing in Metal.
steven-johnson
added a commit
that referenced
this pull request
Jan 17, 2024
ardier
pushed a commit
to ardier/Halide-mutation
that referenced
this pull request
Mar 3, 2024
* adds mappings for f16 variants of halide float math * fix clang format errors * trigger buildbots --------- Co-authored-by: Steven Johnson <srj@google.com>
ardier
pushed a commit
to ardier/Halide-mutation
that referenced
this pull request
Mar 3, 2024
Tweak correctness_float16_t so that it uses one of the transcendal functions (sqrt) that were missing in Metal.
1 task
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Outdated suggestions cannot be applied.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.
Currently you can define float16 variables in a function with a metal schedule, but if codegen emits any float16 math intrincs, the metal compilation will fail because only f32 variants were defined. This adds f16 variants.