2011/1/18 Matthew Knepley <knepley at gmail.com> > This has the EXACT PROBLEM I mailed about last time. Look at your > file > > ksp_poisson2d_256_4_with_cuda_view.txt > > It is only 6K. All the others are 14K. It has no information from ksp_view. >
That is because it is not converging and was interruped. Compare to the smaller problem size where CUDA and non-CUDA converge identically. -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20110120/6fe6a368/attachment.html>
