langou
langou
All this is very interesting. Some thoughts: (1) we should keep these matrices in our torture test suite, it seems like they trigger some difficulties; (2) maybe getting the bidiagonal...
This might be related to #344
@oamarques wrote: Confirmed for N=26 (see below) with gfortran and ifort. Note that only S(1) is the same in both cases. It looks like some kind of splitting/scaling is not...
Hi Keita, 0.0009765625 = 1 / 1024 = 2^(-10) SFAC is used in the test at https://github.com/Reference-LAPACK/lapack/blob/7d90a679e44d0d2add75697923a59b2c15435011/BLAS/TESTING/cblat1.f#L643 > Since we use BLAS with different (heterogeneous) platforms and runtimes, it will...
Hi @hokb and Window users in general, we are looking for some Windows users who can help with this attempt of a PR. @weslleyspereira is trying to get CI to...
Yes, the best solution, in my opinion, is to use LAPACKE. If there are issues with LAPACKE, let us know. LAPACKE should work for great for C and C++. If...
Thanks @yurivict. I take from your message "The treekin port was fixed by patching . . ." that the issue is resolved. This is a good news. I am closing...
@iyamazaki, can you have a look?
Excellent Ichi (@iyamazaki). Thanks for the fix. @yhfeng: Thanks for the bug report. Please let us know if further problem. I am closing the issue.
Hi all, I am reopening the issue. (1) The comments of the `INFO` parameter in DSYTRF_AA are not correct. (2) We are currently discussing the meaning of `INFO > 0`...