| Thread Previous • Date Previous • Date Next • Thread Next |
Dag Lindbo wrote:
Problem solved: We forgot the -DNDEBUG flag to the compiler (which means uBLAS runs full bounds and conformity checking etc). Best to keep to the 'pkg-config' way of compiling :) Sorry. Dag PS. Is NDEBUG defined by some header which happens to get included when the LinearPDE is used? (this seems strange to me, but I really would like to understand the different behaviors).
No. DOLFIN is compiled using -DNDEBUG, therefore when using LinearPDE from the DOLFIN library you don't see the slow down. uBLAS matrices are templated, that's why you see a slow down when not using -DNDEBUG and applying boundary conditions yourself.
Garth
Hi! We've run into what appears to be a strange bug with applying boundary conditions to uBlasMatrix. We implement assembly and boundary condition application (BoundaryCondition::apply()) in two different ways: manually and by LinearPDE. In the manual case (which replicates the LinearPDE implementation), we see an extreme slowdown in what looks to be the uBlasMatrix::ident() function. If we use the LinearPDE implementation the performance is as expected. I'm attaching a simple test case which demonstrates the issue. Has anyone experienced a similar issue or has an idea what goes wrong? This is with DOLFIN 0.7.2 (and should also appear with 0.7.1). Thanks, Johan _______________________________________________ DOLFIN-dev mailing list DOLFIN-dev@xxxxxxxxxx http://www.fenics.org/mailman/listinfo/dolfin-dev_______________________________________________ DOLFIN-dev mailing list DOLFIN-dev@xxxxxxxxxx http://www.fenics.org/mailman/listinfo/dolfin-dev
| Thread Previous • Date Previous • Date Next • Thread Next |
