You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Issue: When upgrading libblas to the new build number, I can not get the prior precision of my BLAS computation as before. The convergence criteria are unchanged. I have attached a script (requiring pyemma from Conda-Forge) to produce the error.
There was an issue in the past, where a certain build causes wrong results on Intel CPU's xref: markovmodel/PyEMMA#1426
The solution back then was to mark this particular build as broken on CF, as the bug was fixed in a new version (e.g. the function dgemm was broken). But I do not know, if this is a similar issue now.
Please tell me, if I should file this one upstream.
Issue: When upgrading libblas to the new build number, I can not get the prior precision of my BLAS computation as before. The convergence criteria are unchanged. I have attached a script (requiring pyemma from Conda-Forge) to produce the error.
There was an issue in the past, where a certain build causes wrong results on Intel CPU's xref: markovmodel/PyEMMA#1426
The solution back then was to mark this particular build as broken on CF, as the bug was fixed in a new version (e.g. the function dgemm was broken). But I do not know, if this is a similar issue now.
Please tell me, if I should file this one upstream.
failing env:
Details
The text was updated successfully, but these errors were encountered: