-
Notifications
You must be signed in to change notification settings - Fork 1.5k
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
AVX2: dgemm on HASWELL leads to invalid results (low accuracy) #4747
Comments
Curious, the current DGEMM kernel for Haswell has been in use for years. Have you only started testing/using OpenBLAS recently, or did something else (like the default compiler version or options) change in your setup lately ? |
We were compiling without DYNAMIC_TARGET thus selecting NEHALEM arch as the minimal supported target. All the opened bugs are due to a more active cross-checking of Windows vs Linux results on our test base after the GCC 8 to GCC 11 Linux migration. Note: all these corner-case bugs might or might not be valid at some points; eg. in that case, EPSILON might be acceptable in most codebases. |
Symbolic result is e-7 off, so both code paths do very well at rounding artifacts |
After upgrading Scilab OpenBLAS build, we detected a poor precision result impacting
dgemm
when using AVX2 kernels.With OpenBLAS 0.3.27:
With refBLAS 3.12.0 :
test code
The text was updated successfully, but these errors were encountered: