-
Notifications
You must be signed in to change notification settings - Fork 79
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
testing: raise ddot bottom_up test to linalg.generic #3678
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #3678 +/- ##
=======================================
Coverage 91.30% 91.30%
=======================================
Files 466 466
Lines 58357 58357
Branches 5624 5624
=======================================
+ Hits 53281 53283 +2
+ Misses 3628 3627 -1
+ Partials 1448 1447 -1 ☔ View full report in Codecov by Sentry. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Makes sense
%X : memref<128xf64>, | ||
%Y : memref<128xf64>, | ||
%G : memref<f64> | ||
%X: memref<128xf64> {llvm.noalias}, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
why do we need this attribute here but not other tests?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Technically it's always needed it, and we make some incorrect assumptions of non-aliasing in our lowering. In the riscv-paper-experiments repo, we have them on all the kernels, as LLVM doesn't do the same optimisations without these annotations.
Last one of the f64, we didn't include it in the paper in the end, but it's worth keeping track of.
Note stacked PR