On May 7, 2011, at 5:01 PM, Jeff Squyres wrote: > example, if you have to pass an MPI datatype down to PARPACK, you'd actually > need to pass MPI_Type_f2c(MPI_INTEGER), or MPI_Type_f2c(MPI_DOUBLE_PRECISION).
Blarg -- I got these backwards. I should have said: > MPI_Type_c2f(MPI_INTEGER), or MPI_Type_c2f(MPI_DOUBLE_PRECISION). Because the MPI_INTEGER and MPI_DOUBLE_PRECISION handles you're using are C handles, and you need to translate them to the equivalent Fortran handles (i.e., the integer handle values you see in mpif.h.). Try it -- cout the return value from MPI_Type_c2f(MPI_INTEGER) and it should match the value you see in mpif.h. -- Jeff Squyres [email protected] For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
