[mpich-discuss] Not able to run HPL Benchmark...

[Jeff Hammond]

Does HPL run in serial?  Can you run cpi on the same number of nodes?

The Graph500 is a much more interested benchmark than HPL.  HPL is
nothing more than a fancy wrapper to DGEMM that causes CPU
manufacturers to overload their chips with vector FPUs and L3 cache
rather than making them more useful for nontrivial problems.  By
running HPL, you perpetuate a 30 year-old myth about scientific
computing that was barely true in the first place.

Jeff

On Mon, Jul 9, 2012 at 7:34 PM, Pavan Balaji <balaji at mcs.anl.gov> wrote:
>
> From the below error message, all I can tell is your application died
> abruptly.
>
>  -- Pavan
>
> On 05/18/2012 02:10 PM, Albert Spade wrote:
>>
>> Hello Everybody,
>>
>> I created a small cluster of 5 machines. One master and 4 compute nodes.
>> I am having Centos 6.2, mpich2-1.4.1p1, lapack-3.2.1-4.el6.i686,
>> atlas-3.8.4-1.el6.i686, atlas-devel-3.8.4-1.el6.i686,
>> blas-devel-3.2.1-4.el6.i686, blas-3.2.1-4.el6.i686 and hpl.
>>
>> I want to run HPL Benchmark and check the performance of my cluster.
>> I followed the procedure for single node given on :
>>
>> http://manyrootsofallevilrants.blogspot.in/2012/04/hpc-beowulf-style-cluster-using-centos.html
>>
>> But after configuring I am not able to get the desired output. Can
>> somebody tell me whats wrong with my configuration?
>>
>> Thanks.
>>
>> The error I am getting is :
>>   ------------------------------------
>>
>> [root at beowulf hpl]# cd bin/Linux_PII_CBLAS/
>>
>> [root at beowulf Linux_PII_CBLAS]# ls
>>
>> HPL.datxhpl
>>
>> [root at beowulf Linux_PII_CBLAS]# mpiexec -n 4 ./xhpl
>>
>> [mpiexec at beowulf.master] control_cb (./pm/pmiserv/pmiserv_cb.c:197):
>> assert (!closed) failed
>>
>> [mpiexec at beowulf.master] HYDT_dmxu_poll_wait_for_event
>> (./tools/demux/demux_poll.c:77): callback returned error status
>>
>> [mpiexec at beowulf.master] HYD_pmci_wait_for_completion
>> (./pm/pmiserv/pmiserv_pmci.c:205): error waiting for event
>>
>> [mpiexec at beowulf.master] main (./ui/mpich/mpiexec.c:437): process
>> manager error waiting for completion
>>
>>
>>
>>
>>
>> _______________________________________________
>> mpich-discuss mailing list     mpich-discuss at mcs.anl.gov
>> To manage subscription options or unsubscribe:
>> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss
>>
>
> --
> Pavan Balaji
> http://www.mcs.anl.gov/~balaji
>
>
> _______________________________________________
> mpich-discuss mailing list     mpich-discuss at mcs.anl.gov
> To manage subscription options or unsubscribe:
> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss



-- 
Jeff Hammond
Argonne Leadership Computing Facility
University of Chicago Computation Institute
jhammond at alcf.anl.gov / (630) 252-5381
http://www.linkedin.com/in/jeffhammond
https://wiki.alcf.anl.gov/parts/index.php/User:Jhammond