(fwd) [mvapich-discuss] FW: MVAPICH2 does not run...
Choudhury, Durga
Durga.Choudhury at drs-ss.com
Tue Jan 31 16:24:30 EST 2006
Hi Wei
Thanks for the response. Please see my answers to your individual
questions below.
-----Original Message-----
From: wei huang [mailto:huanwei at cse.ohio-state.edu]
Sent: Tuesday, January 31, 2006 12:58 PM
To: Choudhury, Durga
Cc: mvapich-discuss at cse.ohio-state.edu
Subject: Re: (fwd) [mvapich-discuss] FW: MVAPICH2 does not run...
Hi,
> 1. We have (and currently do) run Argonne's MPICH2 on our cluster and
it
> works fine. The reason we want to go with the OSU software is that we
> want to take advantage of the uDAPL interface instead of the TCP/IP
that
> we presently use in Argonne's software. However, my original concern
is
> that the OSU software does not even work on Ethernet based TCP/IP
> whereas Argonne's MPICH2 seems to work fine.
We are a bit confused about this. MVAPICH2 current version is
extened from MPICH2-1.0.2p1. So the TCP/IP stack should work wherever
MPICH2 works. Could you please let us know the MPICH2 version you are
using?
We are using both mpich-1.2.7-64bits and mpich2-1.0.2p1-64bits
Also, in your earlier emails you said that you made some changes to get
the code compile on your system. Could you please let us know what
changes
you had made (a patch will be preferred) and send us the
compiling/configure script you are using? Also, as a comparison, would
you
please send us the compiling script that you used for MPICH2.
Unfortunately I don't remember all the details of the change I did
(basically I was hacking with the software until the compiler won't
break). It were mostly very simple things like path names, includepaths
and things like that (although I do realize that including/linking from
the wrong files could be catastrophic, I am pretty sure I did the right
thing.) There was one major change required, however. The "config.guess"
file in mvapich2-0.9.2/src/mpe2/src/slog2sdk/ directory failed to
recognize the platform. I copied the "config.guess" file in
mvapich2-0.9.2/src/mpe/ directory into this directory (which recognized
the platform as mips64-unknown-gnu-linux) for the configure script to
proceed.
> 2. About possible network issues with name resolutions:
> Like I said in the original email, the OSU software works fine on two
> Linux PCs connected back-to-back via Ethernet (i.e. they are not on
any
> network). So I find it strange that our MIPS64 cluster (which ARE
> connected to outside network via front panel Ethernet) would complain
> about name resolution issues.
Again, there may be some problems with your system setup if python
gethostname_ex throws exception. You might want to consult your sysadm
about this issue.
You are probably right; but I am still confused about how this works for
two back-to-back connected Linux PCs (they don't even run the same
Linux; one runs Mandrake and the other runs Fedora Core 4, perhaps even
the kernel versions are different.) And if you could please explain to
me as to why do we care about a name lookup anyway.
Thanks a lot for your help.
Best regards
Durga
Thanks!
Regards,
Wei Huang
774 Dreese Lab, 2015 Neil Ave,
Dept. of Computer Science and Engineering
Ohio State University
OH 43210
Tel: (614)292-8501
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20060131/5f7ddcb4/attachment-0001.html
More information about the mvapich-discuss
mailing list