Piotr Luszczek

Results 5 comments of Piotr Luszczek

I think this issue confuses 32-bit overflow with float-integer conversion. The conversion from 32-bit float to 32-bit integer can be at most off by 256=2^8 because 32-bit floating point can...

Please keep in mind that LAPACK is a long running software library with an established naming scheme. LAPACK also introduces new naming conventions as their needed for clarity. This new...

We don't see this issue on x86 with magma. Maybe @abdelfattah83 seen it with some Grace Hopper builds.

I agree that ScaLAPACK might not be getting the same attention as LAPACK. I'll ask around to see what can be dome to improve the response rate.

Yes, we're preparing one for [SC25](https://sc25.supercomputing.org) in November but it should come this month. @mgates3 might already have the release in the works.