Search found 18 matches

by brom
Wed Sep 05, 2012 8:12 am
Forum: User discussion
Topic: wishlist
Replies: 4
Views: 2432

Re: wishlist

Is magma_zstedx even accelerated? It looks like a complete pass through to the host LAPACK.
by brom
Wed May 02, 2012 2:30 pm
Forum: User discussion
Topic: magma_zgetri_gpu segfaults in OpenMP parallel for
Replies: 14
Views: 10209

Re: magma_zgetri_gpu segfaults in OpenMP parallel for

Thanks Keitat for pointing out the reasons why MAGMA is not thread safe. It's worth noting that the MAGMA was probably "more thread safe" before the CUDA 4.0 changes to GPU binding. In an unrelated project, I was required to completely overhaul my multi-threaded code to account for the multi-GPU cha...
by brom
Wed Apr 04, 2012 4:31 pm
Forum: User discussion
Topic: magma_zgetri_gpu segfaults in OpenMP parallel for
Replies: 14
Views: 10209

Re: magma_zgetri_gpu segfaults in OpenMP parallel for

A developer can correct me if I'm wrong, but I don't think MAGMA is thread safe.
by brom
Tue Apr 03, 2012 6:50 pm
Forum: User discussion
Topic: Tridiagonal Solver
Replies: 3
Views: 2899

Re: Tridiagonal Solver

mgates3 wrote:This is because a tridiagonal solve has O(n) operations on O(n) memory, so is memory bandwidth limited.
This isn't entirely true. Check out NVIDIA's paper on cyclic-reduction algorithms (not in MAGMA).
by brom
Thu Mar 22, 2012 3:23 pm
Forum: User discussion
Topic: dtrsm bug in magma 1.1
Replies: 7
Views: 4849

Re: dtrsm bug in magma 1.1

I've found that almost every MAGMA BLAS kernel takes liberties with reading out of bounds GPU memory when reading into shared memory buffers. On some systems this is OK. However, on others (like yours) it causes errors. Typically this memory won't be accessed so it's not an obvious error, but an err...
by brom
Wed Jan 18, 2012 10:18 am
Forum: User discussion
Topic: syevd performance
Replies: 1
Views: 911

Re: syevd performance

MATLAB uses Intel's MKL which has a special algorithm called successive band reduction for the symmetric eigenvalue factorization step.
by brom
Thu Jun 23, 2011 1:24 pm
Forum: User discussion
Topic: magmablas_dgemv anomalous behavior
Replies: 2
Views: 1704

Re: magmablas_dgemv anomalous behavior

There was a lot of changes to multiGPU in CUDA 4.0 when it came out a few months ago... perhaps MAGMA hasn't updated to support that?
by brom
Wed Mar 02, 2011 10:12 am
Forum: User discussion
Topic: Fremi and Tesla support in one library?
Replies: 2
Views: 1909

Fremi and Tesla support in one library?

Is it possible to compile MAGMA for both Tesla and Fermi support in one library? Right now you must specify the generation in the Makefile. I have a pre-compiled application where I don't know what generation the end user will have. They might have a Fermi, Tesla, or even a G80. I know this is possi...
by brom
Wed Feb 23, 2011 6:58 pm
Forum: User discussion
Topic: magma_dsytrd questions
Replies: 25
Views: 8143

Re: magma_dsytrd questions

OK, thanks. That explains why the GPU performance is so low. Do you think these routines will be in RC4?

Is it also safe to assume the numbers from the paper are from only single-threaded LAPACK? That seems to be the conclusion we have found here.
by brom
Wed Feb 23, 2011 5:35 pm
Forum: User discussion
Topic: magma_dsytrd questions
Replies: 25
Views: 8143

Re: magma_dsytrd questions

So MAGMA's SSYMV and SGEMV aren't the ones from the paper?