On May 12, 2010, at 1:48 PM, Hanjun Kim wrote: > I am working on parallelizing my sequential program using OpenMPI. > Although I got performance speedup using many threads, there was > slowdown on a small number of threads like 4 threads. > I found that it is because getc worked much slower than sequential > version. Does OpenMPI override or wrap getc function?
No. > To find the cause, I added mpi.h to the program and compiled it with > mpicc. There are two versions on the program; > 1. The sequential program without MPI_Init_thread and MPI_Finalize. > 2. The sequential program with MPI_Init_thread and MPI_Finalize. > Version 2 was 4x slower than version 1. I think it is because of slow > getc. Does MPI_Init_thread have some relation with the getc function > call? Not that I'm aware of. Perhaps something about linking in threading libraries changes getc...? -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
