Crashing ibverbs binary of NAMD_2.7b2

From: Philipp M. (philipp.coloc_at_gmx.de)
Date: Wed Feb 10 2010 - 10:58:03 CST

Hi,

besides trying to compile the source code (please reply to my previous post), I also try to run the ibverbs binary.
This is the ++verbose output:

Charmrun> charmrun started...
Charmrun> using ./nodelist as nodesfile
Charmrun> adding client 0: "master0", IP:127.0.0.1
Charmrun> adding client 1: "master0", IP:127.0.0.1
Charmrun> Charmrun = master0, port = 43809
Charmrun> IBVERBS version of charmrun
Charmrun> Sending "0 master0 43809 32619 0" to client 0.
Charmrun> find the node program "/usr/local/testit/NAMD_2.7b2_Linux-x86_64-ibverbs/namd2" at "/data/nvt_run8a" for 0.
Charmrun> Starting rsh master0 -l user /bin/sh -f
Charmrun> remote shell (master0:0) started
Charmrun> Sending "1 master0 43809 32619 0" to client 1.
Charmrun> find the node program "/usr/local/testit/NAMD_2.7b2_Linux-x86_64-ibverbs/namd2" at "/data/nvt_run8a" for 1.
Charmrun> Starting rsh master0 -l user /bin/sh -f
Charmrun> remote shell (master0:1) started
Charmrun> node programs all started
Charmrun remote shell(master0.1)> remote responding...
Charmrun remote shell(master0.0)> remote responding...
Charmrun remote shell(master0.1)> starting node-program...
Charmrun remote shell(master0.1)> rsh phase successful.
Charmrun remote shell(master0.0)> starting node-program...
Charmrun remote shell(master0.0)> rsh phase successful.
Charmrun> Waiting for 0-th client to connect.
Charmrun> error 93620 attaching to node:
Socket closed before recv.

In the charm++ FAQ you find:
"typically means a segmentation fault"

So is this really a bug? Let me repeat that our mpi-compiled version of NAMD_2.7b1 works fine with our infiniband implementation.

Best,
Philipp

-- 
Jetzt kostenlos herunterladen: Internet Explorer 8 und Mozilla Firefox 3.5 -
sicherer, schneller und einfacher! http://portal.gmx.net/de/go/chbrowser

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:55:24 CST