Skip to content
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

Try harder at disabling gemv_n unrolling #1768

Closed
wants to merge 2 commits into from
Closed

Try harder at disabling gemv_n unrolling #1768

wants to merge 2 commits into from

Conversation

martin-frbg
Copy link
Collaborator

for #1748

@martin-frbg
Copy link
Collaborator Author

No longer convinced this is even necessary... there seems to be a small lack of precision caused by the Haswell implementation of gemv_n in dgemv_n_4.c compared to the plain C arm/gemv_n.c but this is hardly the fault of the trmv_U.c caller and shows up only in the (fixed) test case from #1332 as a MAXERR of .2842D-13 instead of .1421D-13 achievable with the plain C version for N=507. By comparison, allowing multithreading in interface/trmv.c increases the error to around .15D-12

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant