[release/9.0-rc2] Replace VectorXx.Exp's edge case fallback with scalar processing #107942
Conversation
This file contains hidden or 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
The better, vectorized fix is more complex and can be done for .NET 10.
ghost
added
the
Contributor
|
Tagging subscribers to this area: @dotnet/area-system-numerics |
Member
|
cc: @jeffhandley, @artl93 |
eiriktsarpalis
approved these changes
Sep 18, 2024
jeffhandley
approved these changes
Sep 18, 2024
jeffhandley
added
Servicing-approved
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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Outdated suggestions cannot be applied.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.
Backport of #107886 to release/9.0-rc2
/cc @stephentoub
Customer Impact
Certain edge case double values passed to Vector64/128/256/512 and TensorPrimitives.Exp produce the wrong result, e.g. negative infinity instead of a value close to 0. There's a special path for handling such edge cases while still vectorizing, and that vectorized result can be wrong. The fix is to handle those edge cases by falling back to scalar processing for the vector containing the edge case.
Regression
Testing
Added new tests for the edge cases.
Risk
Low. It's falling back to the scalar implementation when such values are encountered.