Search found 90 matches

by hartwig anzt
Mon Feb 22, 2016 4:43 pm
Forum: User discussion
Topic: Magma Sparse Example
Replies: 1
Views: 3872

Re: Magma Sparse Example

Dear Costa, I can not exactly reproduce your problem, but have some comments. You use opts.solver_par without initializing most of the fields. E.g. you should set: opts.solver_par.maxiter = 1000; opts.solver_par.rtol = 1e-10; opts.solver_par.rtol = 1e-10; opts.solver_par.restart = 50; An elegant wor...
by hartwig anzt
Sun Oct 18, 2015 9:06 am
Forum: User discussion
Topic: LOBPCG method for Eigen Value Computation.
Replies: 4
Views: 1956

Re: LOBPCG method for Eigen Value Computation.

Dear Nithish, for a plain LOBPCG, use ./testing_dsolver --solver 8 --ev k /path/to/the/matrix If you want to use a preconditioner (faster solver convergence in many cases) type ./testing_dsolver --solver 8 --ev k --precond x /path/to/the/matrix where x = 1 is Jacobi, x = 2 is ILU. Let me know whethe...
by hartwig anzt
Tue Sep 15, 2015 7:30 am
Forum: User discussion
Topic: Solving with multiple RHS
Replies: 8
Views: 3307

Re: Solving with multiple RHS

Maybe you can write out one sample matrix, such that I can try to figure out what is happening? If yes, just pass me the URL.

Thanks, Hartwig
by hartwig anzt
Tue Sep 15, 2015 4:52 am
Forum: User discussion
Topic: Solving with multiple RHS
Replies: 8
Views: 3307

Re: Solving with multiple RHS

Badlishah,

this error typically occurs if the linear system is not positive definite, which is needed for using CG. In particular, the system needs to be positive ( A=A' ) and all eigenvalues of the system need to be positive. Is this true for your system? What is the origin?

Thanks, Hartwig
by hartwig anzt
Mon Sep 14, 2015 11:58 pm
Forum: User discussion
Topic: Solving with multiple RHS
Replies: 8
Views: 3307

Re: Solving with multiple RHS

Badlishah,

indeed, not all solvers support block-RHS at this point. Please try with preconditioned Conjugate Gradient (called zbpcg).

Thanks, Hartwig
by hartwig anzt
Sun Sep 06, 2015 12:59 am
Forum: User discussion
Topic: Solving with multiple RHS
Replies: 8
Views: 3307

Re: Solving with multiple RHS

Badlishah, at the current state, I would suggest to transform the sparse RHS-matrix into a dense matrix, such that you have a block of vectors (dense matrix) RHS. magma_zmconvert( B, &B2, Magma_CSR, Magma_DENSE, queue ); Then, you use an initial guess for the block solution (X) also as dense matrix....
by hartwig anzt
Tue Sep 01, 2015 11:26 am
Forum: User discussion
Topic: CSC To CSR Matrix Conversion
Replies: 5
Views: 2309

Re: CSC To CSR Matrix Conversion

Badlishah, There is no function for this at this point (all MAGMA sparse is using 0-based indexing). At this point, I would do this manually on the CPU: for( magma_int_t i=0; i<A.num_rows+1; i++) { A.row += 1; } for( magma_int_t i=0; i<A.nnz; i++) { A.col += 1; } Please let me know if you have probl...
by hartwig anzt
Fri Aug 14, 2015 9:02 am
Forum: User discussion
Topic: Help please to get CUDA 6.5 working on Ubuntu 12.04
Replies: 3
Views: 1303

Re: Help please to get CUDA 6.5 working on Ubuntu 12.04

John,
did you consider updating your ubuntu version? this might be the easier way to go.
Hartwig
by hartwig anzt
Mon Aug 03, 2015 10:21 am
Forum: User discussion
Topic: Magma version of BLAS routine ZSBMV
Replies: 4
Views: 903

Re: Magma version of BLAS routine ZSBMV

John,

now I see what you want: element-wise multiplication of a vector. And you found a workaround using a sparse matrix. I promise, once you have the code running on the GPU using MAGMA, I'll write a function doing this quickly. This is very easy on a GPU.

Hartwig
by hartwig anzt
Mon Aug 03, 2015 8:25 am
Forum: User discussion
Topic: Magma version of BLAS routine ZSBMV
Replies: 4
Views: 903

Re: Magma version of BLAS routine ZSBMV

John, if I do understand you correct, you want to multiply a sparse matrix with a vector - is this correct? And the sparse matrix has a nice diagonal property? If so, I would recommend just using a default sparse matrix vector (SpMV) routine. These are typically highly optimized, and will sure satis...