Search found 90 matches

by hartwig anzt
Sat Aug 01, 2015 9:17 pm
Forum: User discussion
Topic: Seeking advice on choice of sparse matrix format
Replies: 4
Views: 1890

Re: Seeking advice on choice of sparse matrix format

John, are you working on your personal Desktop computer or do you have access to some cluster? In the second case, I could probably help getting the code running - if granted access. For the CPU side, Intel's MKL provides some sparse support. Unfortunately, I have to admit that the FORTRAN support i...
by hartwig anzt
Sat Aug 01, 2015 4:53 pm
Forum: User discussion
Topic: Seeking advice on choice of sparse matrix format
Replies: 4
Views: 1890

Re: Seeking advice on choice of sparse matrix format

Dear John, a 4000x4000 matrix is still relatively small. Sparse iterative systems typically target problems with more than 100,000 unknowns. So it might be an option to handle this system as dense. However, if you want, you can handle it as sparse. The easiest way is to convert it to csr, and then u...
by hartwig anzt
Tue Jul 28, 2015 9:57 am
Forum: User discussion
Topic: JDQZ method for the generalized eigenvalue problem
Replies: 2
Views: 739

Re: JDQZ method for the generalized eigenvalue problem

Dear John, thank you for pointing us to this! Generally, we would be very interested in having this - in particular for general eigenvalue problems. Unfortunately, we only do have very limited resources, so I can't promise we'll add this method within the next weeks. What is the application you woul...
by hartwig anzt
Wed Jul 22, 2015 9:33 pm
Forum: User discussion
Topic: Test routines for the sparse matrix methods in 1.6.2
Replies: 5
Views: 1671

Re: Test routines for the sparse matrix methods in 1.6.2

John, I can completely understand your concern. I found a few reference for upgrading to 6.5 on ubuntu 12.04, so you may give it a try: http://jasonjuang.blogspot.com/2013/09/how-to-install-cuda-55-on-ubuntu-1204.html http://ivonneleonor.blogspot.com/2014/09/first-of-all-you-have-to-download-from.ht...
by hartwig anzt
Wed Jul 22, 2015 3:25 pm
Forum: User discussion
Topic: Test routines for the sparse matrix methods in 1.6.2
Replies: 5
Views: 1671

Re: Test routines for the sparse matrix methods in 1.6.2

Dear John, to generate the sparse library, just enter the sparse-iter folder and follow the instructions. The building process is very similar to the dense MAGMA. MAGMA is in general backward-compatible. CUDA 4 is pretty old, if you can, I would recommend getting a newer version. The newer versions ...
by hartwig anzt
Mon Jul 13, 2015 3:58 pm
Forum: User discussion
Topic: DGMRES gone?
Replies: 2
Views: 455

Re: DGMRES gone?

Dear Josef, you are right, dgmres is still listed in the header, but does no longer exist. dfgmres is a flexible GMRES implementation, this allows for changing preconditioners. If you want a plain (unpreconditioned) GMRES, using dfgmres without preconditioner is almost equivalent. The performance di...
by hartwig anzt
Wed Jun 17, 2015 2:01 pm
Forum: User discussion
Topic: What's the role of restart in PGMRES?
Replies: 1
Views: 726

Re: What's the role of restart in PGMRES?

Dear Harshad, the restart parameter determines the dimension of the Krylov space before restarting. Please take a look at the literature for more theoretical background: http://www-users.cs.umn.edu/~saad/IterMethBook_2ndEd.pdf Small restart values result in low memory cost, faster loop execution (li...
by hartwig anzt
Wed Jun 17, 2015 1:38 pm
Forum: User discussion
Topic: What's wrong with the ILU sparse preconditioner?
Replies: 1
Views: 1038

Re: What's wrong with the ILU sparse preconditioner?

Dear Harshad, The test problem you are addressing (Laplace 2D of size 40,000) is a very structured, very sparse problem. In particular, it contains only about 5 nnz/row. This is why operations like SpMV or Jacobi-preconditioning are very fast. They can be parallelized by rows. For the ILU preconditi...
by hartwig anzt
Thu Apr 23, 2015 9:13 am
Forum: User discussion
Topic: Numerical Errors
Replies: 3
Views: 1458

Re: Numerical Errors

Can you please be a bit more specific - which tests fail? which routines do you use? Are you talking about dense or sparse solvers? Which test matrices do you use?

Thanks, Hartwig
by hartwig anzt
Fri Apr 03, 2015 4:51 pm
Forum: User discussion
Topic: Sparse Direct Solver
Replies: 3
Views: 1548

Re: Sparse Direct Solver

The LU is an incomplete LU (ILU) this can be used as a preconditioner for iterative solvers.