[Beowulf] Measuring port to port latency

[Tom Elken]

 > -----Original Message-----
> From: beowulf-bounces at beowulf.org 
> [mailto:beowulf-bounces at beowulf.org] On Behalf Of Kozin, I (Igor)

> The Network-Based Computing Laboratory at the Ohio State 
> University used to host a number of benchmarking programs 
> including the latency test.
> Unfortunately I do not have the link.

http://mvapich.cse.ohio-state.edu/benchmarks/

I believe that these OSU MPI benchmarks together with the IMB (Intel MPI
Benchmarks, formerly Pallas) are the most commonly used ping-pong (=
latency) tests that people use for MPI.

> I found that 
> PathScale's link checker (derived from an OSU benchmark) is 
> very handy. Perhaps you can find the link in the archive. 

PathScale's InfiniBand/MPI technology is now part of QLogic.

>From the Archive:
    cse.ucdavis.edu/~bill/mpi_nxnlatbw.c
	Redistribution and use in source and binary forms, with or
	without ... Redistributions in binary form must reproduce the
	above copyright notice, this ...
    http://cse.ucdavis.edu/~bill/mpi_nxnlatbw.c [#1 on Yahoo!]

Info about this program is available in the QLogic InfiniPath 2.1 User
Guide:
http://download.qlogic.com/manual/62612/IB6054601-00E.pdf
mpi_nxnlatbw is run as part of the ipath_checkout tool (p. C-38) that
checks whether all parts of the network are installed correctly and
performing as expected.  Or it can be run separatly.

-Tom

> Also check the archive on the issues with measuring latency.
> Best,
> Igor 
> 
> -----Original Message-----
> From: beowulf-bounces at beowulf.org [mailto:beowulf-bounces at beowulf.org]
> On Behalf Of Carsten Aulbert
> Sent: 20 September 2007 12:12
> To: Beowulf Mailing List
> Subject: [Beowulf] Measuring port to port latency
> 
> Hi,
> 
> since the rush of emails seems to be over right now, I would 
> like to start another thread. How are you measuring port to 
> port latency in your network easily?
> 
> So far, I have found a small MPI-pingpong code at
> 
> http://cluster.fisica.unam.mx/curso/mpi_intro/para_pingpong.html
> 
> as well as netpipe at
> 
> http://www.scl.ameslab.gov/Projects/Netpipe/
> 
> They look fine on their own, but since I'm just starting, any 
> opinions?
> 
> Cheers
> 
> Carsten
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org To change your 
> subscription (digest mode or unsubscribe) visit 
> http://www.beowulf.org/mailman/listinfo/beowulf
> 
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org To change your 
> subscription (digest mode or unsubscribe) visit 
> http://www.beowulf.org/mailman/listinfo/beowulf
>