[mvapich-discuss] is it bug when a large connection, large cluster, is used on mvapich 0.9.9-2007-04-25 daily verion?

[Sayantan Sur]

Hi,

??? wrote:
> Hello
> When I ran HPL on my cluster sysem which is consist of linux blade 
> system , it die with a strange error.
> Under using 640 process then it works well but HPL die above 640 .
> The error is blow.
> bash : line 1 : 18891 Segmentation fault /usr/bin/env 
> LD_LIBRARY_PATH=... ... ...VIA_... ... ... ./xhpl
> -------------------------- --------- -----
> LD_LIBRARY_PATH environment value IB environment program
>
> Because i don't know about this error why it occurs so i want to help
>
> and I used IB driver is ofed 1.2 rc 2 and MVAPICH 0.9.9 2007- 04-25 
> version.
>
> Have Any idea about this problem?? t

Thanks for trying out MVAPICH-0.9.9. Sorry that you faced a segmentation 
fault above 640 processes. To better diagnose the error and its causes, 
could you please send us the back-trace of the segmentation fault? You 
could do that by compiling both MVAPICH and HPL with "-g" in the CFLAGS 
(for MVAPICH it should be placed in make.mvapich.gen2). Then enable core 
dumps (ulimit -c unlimited) in your shell. Start gdb on the core file 
and say "bt". That should give us some more information about where the 
application is segfaulting.

Thanks,
Sayantan.

>
> Thanks.
>
> -jk
> ------------------------------------------------------------------------
>
> _______________________________________________
> mvapich-discuss mailing list
> mvapich-discuss at cse.ohio-state.edu
> http://mail.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss
>   


-- 
http://www.cse.ohio-state.edu/~surs