Search found 41 matches

by mateo70
Thu Apr 14, 2011 1:22 am
Forum: User discussion
Topic: Timing of dgetrs_gpu and zgetrs_gpu
Replies: 6
Views: 3009

Re: Timing of dgetrs_gpu and zgetrs_gpu

I just added the switch between the two functions trsm and trsv
We still have to implement a magma version.

Mathieu
by mateo70
Thu Apr 14, 2011 1:19 am
Forum: User discussion
Topic: Error compiling on OS X
Replies: 4
Views: 2541

Re: Error compiling on OS X

Hi,

Can you tell us if you still have the problem with the new RC5. The fortran interface is now implemented, before it was only a prototype.

Thanks,
Mathieu
by mateo70
Wed Apr 06, 2011 2:46 pm
Forum: User discussion
Topic: Fortran Subarrays on GPU in RC4
Replies: 7
Views: 4620

Re: Fortran Subarrays on GPU in RC4

Yes, I will do that.
Hopefully I looked at this post before to do it :).

Mathieu
by mateo70
Wed Apr 06, 2011 2:43 pm
Forum: User discussion
Topic: Timing of dgetrs_gpu and zgetrs_gpu
Replies: 6
Views: 3009

Re: Timing of dgetrs_gpu and zgetrs_gpu

Yes, we discussed about that with Stan, but apparently the version actually in the code was the one giving the best performance.
But I will add the test case for one RHS.

Mathieu
by mateo70
Tue Apr 05, 2011 6:50 pm
Forum: User discussion
Topic: Transposed cases with zgetrs and cgetrs
Replies: 2
Views: 2592

Re: Transposed cases with zgetrs and cgetrs

John,

thanks for the information. I havn't seen it when it has been commited. I just corrected it and will be integrated in the next release. (I hope by the end of the week)

Mathieu
by mateo70
Tue Apr 05, 2011 6:44 pm
Forum: User discussion
Topic: Trouble in running test programs
Replies: 16
Views: 10607

Re: Trouble in running test programs

Hi Yu,

The problem was that he was linking with two different libcublas.so, and they were a mix between them.
Try to do a ldd on the binary to check that you are using the correct library.

Mathieu
by mateo70
Tue Apr 05, 2011 6:41 pm
Forum: User discussion
Topic: Error compiling on OS X
Replies: 4
Views: 2541

Re: Error compiling on OS X

Peter,

I'm sorry, we are not supporting Mac OS for now and it's not in our future plan. But we are open to integrate patches from users if they want to do the port on Mac as it has been done (or in progress) for Windows.

Mathieu
by mateo70
Mon Mar 21, 2011 12:14 pm
Forum: User discussion
Topic: Performance of data transfers to GPU
Replies: 10
Views: 8143

Re: Performance of data transfers to GPU

John, just to know, do you really need to do all the solve separatly ? can't you just call getrs_gpu with all the rhs ? Otherwise these results looks interesting, but I will try to fix that for next release since we have a function to do the swap directly on GPU without the need to transfer the RHS ...
by mateo70
Mon Mar 21, 2011 12:07 pm
Forum: User discussion
Topic: Fortran Subarrays on GPU in RC4
Replies: 7
Views: 4620

Re: Fortran Subarrays on GPU in RC4

Thanks john,

that's close to what we planned to include in the next release with the fortran interface. I'm just busy with other projects right now, so I don't have a date for this final release.
The prototype we were thinking about is:

Code: Select all

 magma_[zcds]offset( NewPtr, OldPtr, LDA, I, J) 
Mathieu
by mateo70
Tue Mar 15, 2011 5:20 pm
Forum: User discussion
Topic: Changes in RC4
Replies: 5
Views: 3080

Re: Changes in RC4

Yes , sorry I forgot this change, since I didn't do it. I have to check why it doesn't work with the size of a pointer (integer kind=8 on 64bits system or integer kind=4 for 32bits system) which should be the right choice for this. And we change it to this value because some compiler were not happy ...