[mpich-discuss] Flash IO Benchmark using mpich2 and pvfs2

[Shawn Kim]

Hi,

I've tried to implement the entire trace from the application program
through pvfs2 clients to servers and did it. (I tested a simple application
program and got full paths)
Now, I want to trace MPI_File_write_all from the pnetcdf used in Flash IO
(argonne version) to the PVFS2.

I found that pnetcdf issued 50 MPI_File_write_all() calls and hdf5 did 32
MPI_File_write_at_all()'s.
In addition, I knew that, PVFS_sys_write() calls didn't seem to be passed to
the PVFS_sys_io() in sys-io.sm. (I modified the entire IO stack to trace IO
related function, NOT using MPE)


1. MPI_File_write_all (mpich2-1.0.8/src/mpi/romio/mpi-io/write_all.c)
2. MPIOI_File_write_all (mpich2-1.0.8/src/mpi/romio/mpi-io/write_all.c)
3. ADIO_WriteStridedColl
4. ADIOI_GEN_WriteStridedColl
(mpich2-1.0.8/src/mpi/romio/adio/common/ad_write_coll.c)
-------------------------------------------------------------------------------------------------------------------------------------
>From here, I can't trace it
5. ADIO_WriteContig (ad_pvfs2_write.c)
6. ADIOI_PVFS2_WriteContig
7. PVFS_sys_write
====================================
8. PVFS_sys_io (sys-io.sm)

Is there anybody who knows why these call sequences don't work? If anybody
ran Flash IO before, is the result correct?
Or, am I missing something?

Thank you

- Best regards,
Shawn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20100310/d85c1a58/attachment.htm>