[mvapich-discuss] MVAPICH2-0.9.8 internal errors

David Gunter dog at lanl.gov
Thu Mar 22 11:31:40 EDT 2007


I have recompiled mvapich2 without using the --enable-debuginfo flag  
and the problem has gone away.  However, I wish to have debuginfo  
available to our TotalView users so hopefully this can be resolved.

Here is the configuration that generates the error message I saw  
previously:

./configure --prefix=/opt/mvapich2/mvapich2-0.9.8-gcc/ib --with- 
openib=/usr/local/ofed --enable-romio --with-file-system=ufs+nfs 
+panfs --enable-sharedlibs=gcc --enable-debuginfo --enable-fast -- 
with-mpe

OFED is ofed-1.1

CFLAGS=-D_X86_64_ -D_SMP_ -DUSE_HEADER_CACHING  -DONE_SIDED - 
DMPID_USE_SEQUENCE_NUMBERS  -I/usr/local/ofed/include -O2   - 
D_SHMEM_COLL_

CC=/usr/bin/gcc
CXX=/usr/bin/g++
FC=/usr/bin/gfortran
F77=/usr/bin/gfortran
F90=/usr/bin/gfortran

The test was run on 32 process, 16 process, 8 process and 4 process -  
all of which generated this error message.

Thanks,
david



--
David Gunter
HPC-4: HPC Environments: Parallel Tools Team
Los Alamos National Laboratory


On Mar 21, 2007, at 4:50 PM, wei huang wrote:

> Hi,
>
> Thanks for letting us know this problem.
>
> We will try to reproduce the problem on our cluster. To help us on  
> looking
> into this problem, would you please let us know the following:
>
> 1) The exact CFLAGS you are using when configuring mvapich2 (are  
> you using
> default scripts provided by us?)
> 2) Any runtime environment variables you have setup up?
> 3) On how many nodes do you run the test?
>
> 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
>
>
> On Wed, 21 Mar 2007, David Gunter wrote:
>
>> I have built mvapich2 for an OFED-based IB Opteron cluster.  When
>> running the Intel MPI Benchmarks (IMB3) I keep seeing the following
>> errors messages in many spots, although the tests run to completion:
>>
>> Internal error: communicator is already on free list
>>
>> What is this referring to?
>>
>> Thanks.
>> --david
>>
>> --
>> David Gunter
>> HPC-4: HPC Environments: Parallel Tools Team
>> Los Alamos National Laboratory
>>
>>
>>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20070322/db7870ba/attachment-0001.html


More information about the mvapich-discuss mailing list