[Topi][Relay] Support for FP16 ERF on CPU. #11413
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
|
@anwang2009 can you take a look at this PR? |
|
Is the default behavior that FP16 ERF will fallback to FP32 cast for the ERF call? Basically wondering in what cases the different fallbacks are used? |
|
There's different handling for |
AndrewZhaoLuo
approved these changes
May 23, 2022
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Outdated suggestions cannot be applied.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.
X86 targets don't have an FP16 erf intrinsic. This causes compilation to fail when such an op is encountered. This PR adds two cases to avoid this failure. I've added an FP16 ERF legalization pass that will just cast to FP32 for the ERF call. I've also extended fast_erf to support FP16 directly. This does lose some precision compared to doing FP32 wrapping around ERF, but since the desire is to be faster I think it's acceptable behavior.