[mpich-discuss] MPI_Send/Recv vs MPI_un/pack performance

Roberto Fichera kernel at tekno-soft.it
Thu Jul 31 12:18:52 CDT 2008


Hi All on the list,

My apologies if this argument was already well discussed many times 
before in the list, but playing around the
MPI_Send() and MPI_Recv() I end up on trying to optimize both transfert 
and receive parts for our
serialization/deserialization functions. So looking around I decide to 
use the MPI_Pack() and MPI_Unpack() into
an already allocated buffer and send/receive it in one "big shot". After 
getting it working I finally created a test program
for profiling the performances in both cases. My results was quite 
stunning, working with various sizes in Mb of serialized
data, the MPI_Send() is ~2 to 4 times faster than paired MPI_Pack() & 
MPI_Send()!!! So, my understanding for the
MPI_Pack() is that it can be usefull for packing and unpacking data to 
send/receive for performance reasons, maybe
do I'm missing something (o.s. tuning)?

I'm using mpich2 v1.0.7 with channel ch3:socket on Fedora 8 x86_64.

Best regards,
Roberto Fichera.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: char.png
Type: image/png
Size: 13541 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20080731/6a38b498/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: double.png
Type: image/png
Size: 13402 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20080731/6a38b498/attachment-0001.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: int.png
Type: image/png
Size: 12966 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20080731/6a38b498/attachment-0002.png>


More information about the mpich-discuss mailing list