math_brute_force: treat reciprocal as unary function #2281
+62
−120
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
bashbaug
requested changes
Feb 18, 2025
Treat reciprocal as a unary function, instead of handling it through the binary function testing mechanism and special-casing it there. This addresses two shortcomings of the previous implementation: - Testing took significantly longer as the entire input domain was tested many times (e.g. fp16 reciprocal has only 2^16 possible input values, but binary function testing iterates over 2^16 * 2^16 input values). - The reciprocal test kernel was identical to the divide kernel. Thus the device compiler would see a regular divide operation instead of a reciprocal operation and would be unlikely to emit a specialized reciprocal sequence. This reverts all of the changes in binary_operator*.cpp made by bcfa1f7 ("Added corrections to re-enable reciprocal test in math_brute_force suite for relaxed math mode (KhronosGroup#2221)", 2025-02-04). Signed-off-by: Sven van Haastregt <[email protected]>
svenvh
force-pushed
the
math-recip-unary
branch
from
a9a4efa to
c33e9b0
Compare
|
Works fine with my nvidia and intel devices. |
bashbaug
approved these changes
Feb 19, 2025
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Outdated suggestions cannot be applied.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.
Treat reciprocal as a unary function, instead of handling it through the binary function testing mechanism and special-casing it there.
This addresses two shortcomings of the previous implementation:
Testing took significantly longer as the entire input domain was tested many times (e.g. fp16 reciprocal has only 2^16 possible input values, but binary function testing iterates over 2^16 * 2^16 input values).
The reciprocal test kernel was identical to the divide kernel. Thus the device compiler would see a regular divide operation instead of a reciprocal operation and would be unlikely to emit a specialized reciprocal sequence.
This reverts all of the changes in binary_operator*.cpp made by bcfa1f7 ("Added corrections to re-enable reciprocal test in math_brute_force suite for relaxed math mode (#2221)", 2025-02-04).