Search found 90 matches

by hartwig anzt
Thu Oct 05, 2017 4:37 pm
Forum: User discussion
Topic: Efficient computation (solve) of Ax = b when A is symmetric
Replies: 3
Views: 935

Re: Efficient computation (solve) of Ax = b when A is symme

Dear Klaus, MAGMA-sparse contains iterative Krylov solvers that exploit symmetry information, i.e. the CG method (potentially preconditioned with Jacobi or Incomplete Cholesky). This combination should be able to solve your problem with satisfying efficiency. Most iterative solvers are based on spar...
by hartwig anzt
Mon Oct 02, 2017 3:22 am
Forum: User discussion
Topic: Solver parameters for eigenvalue routines
Replies: 7
Views: 3191

Re: Solver parameters for eigenvalue routines

Dear Olafur, LOBPCG is an iterative eigenvalue solver that can be used to compute a certain number of the largest/smallest eigenvalues. It is a Krylov-based method, and you can add a preconditioner to it. Also the stopping criterion and the total number of iterations are valid. In other words, a set...
by hartwig anzt
Fri Aug 25, 2017 9:33 am
Forum: User discussion
Topic: Exact final residual: -nan/ info= -202
Replies: 3
Views: 1376

Re: Exact final residual: -nan/ info= -202

Dear Ham,

you try to solve a system Ax = b with b=0. Without using any HPC software, I can tell you the solution is x=0.

For your other question: Yes, "dopts.precond_par.solver = Magma_NONE" turns off the preconditioner.


I hope this helps, please feel free to ask other questions. Thanks, Hartwig
by hartwig anzt
Tue Jun 06, 2017 2:55 am
Forum: User discussion
Topic: magma_dcsrset_gpu missing declaration
Replies: 2
Views: 1323

Re: magma_dcsrset_gpu missing declaration

Thanks, Tom, for pointing this out!
I will make sure to fix it in the next release.
Thanks, Hartwig
by hartwig anzt
Thu May 11, 2017 2:03 am
Forum: User discussion
Topic: Magma crashes when i want to solve One milion matrix
Replies: 1
Views: 1157

Re: Magma crashes when i want to solve One milion matrix

Dear leo777, as you point out that the code works for some matrices, and I don't have a matrix at hand where it does not work, it is a bit difficult for me to identify the problem. Maybe you could provide me access to the problem where the whole thing crashes? (Google-Drive, Dropbox, fileshare) Alte...
by hartwig anzt
Fri Apr 21, 2017 11:03 am
Forum: User discussion
Topic: is magma_ssymv availlable for sparse matrices?
Replies: 1
Views: 1107

Re: is magma_ssymv availlable for sparse matrices?

Sorry, there is no symmetric SpMV at this point. You are welcome to contribute one. Otherwise you will have to stick to the non-symmetric kernel. I however doubt you would see large performance benefits from exploiting symmetry, this is very difficult. Please let me know whether you have further que...
by hartwig anzt
Mon Mar 06, 2017 7:56 pm
Forum: User discussion
Topic: magma sparse solver has different answer compare then MKL
Replies: 2
Views: 1886

Re: magma sparse solver has different answer compare then MK

Dear AlarmChang, maybe you use MKL in the wrong fashion? From the last row, you can see right away that the MKL computation has quite some rounding.... 1/5 = 0.2 the last time I checked... Also, if I run the system through matlab, I get very much the same result, see below. Let me know if you have f...
by hartwig anzt
Tue Feb 14, 2017 1:38 pm
Forum: User discussion
Topic: Basic code for SpMV on GPU
Replies: 2
Views: 1924

Re: Basic code for SpMV on GPU

Dear Akrem Benatia, assuming you have your matrix of size n in the CSR-array *row *col *val on the CPU, you can pass this matrix to MAGMA-sparse via magma_d_matrix A={Magma_CSR}, magma_zcsrset( n, n, row, col, val &A, queue ); Then, you may want to initialize vectors on the GPU and copy the system t...
by hartwig anzt
Mon Feb 13, 2017 10:41 am
Forum: User discussion
Topic: MAGMA use for steady CFD simualtions
Replies: 2
Views: 1491

Re: MAGMA use for steady CFD simualtions

Dear "dreamer", thank you for pointing out this issue. I appreciate attaching code that I can follow what is happening. The increased memory usage comes from the fact that the preconditioner is not deallocated after being applied. This means, the ILU factorization stays in the GPU memory, and when t...
by hartwig anzt
Sun Jan 15, 2017 10:23 am
Forum: User discussion
Topic: Magma change storage_type error in GPU
Replies: 4
Views: 2099

Re: Magma change storage_type error in GPU

Mike, this looks good! A few comments, also as answer to your questions. The error handling works the following way: If any of the function returns an error, the "CHECK" will catch this, and proceed at the "cleanup" at the end of the routine. The intention is is free all memory allocated, and return...