Hi David,
> Thank you for your detailed reply. I've expressed my benchmarking
> results in terms of ns/day to allow you to put my figures in context.
> First of all I should mention that we have two M2050 gpus installed in
> each of our compute nodes. Here are my results (expressed in ns/day)
> for the PME/Cellulose_production_NPT benchmarking case:
>
> Conventional (cpu) results ns/day
> 8 cores (1 node) 0.36
> 16 cores (2 nodes) 0.67
>
> GPU results
> 1 1.82
> 2 (1 node) 2.55
> 4 (2 nodes) 3.55
>
> Do these results make sense? It would appear that the simulation
> performed using 4 gpus (2 nodes) is roughly 5 times faster than the
> corresponding conventional case over 16 cores.
Here are my reference numbers for C2070 (totally equivalent to M2050 in
terms of performance) for this benchmark for 2GPUs per node on a QDR IB
system (
http://ambermd.org/gpus/benchmarks.htm#Benchmarks):
Cellulose NPT (2 GPU per node QDR IB)
C2070 ns/day
1 2.02
2 2.91
4 3.36
So your 1 and 2 GPU numbers are a little on the slow side while your 4 GPU
number is better. My guess would be that you have a good QDR IB card in an
x16 slot. In my case the two GPUs in each node are in x16 slots but the IB
card is in an x4 slot.
The difference in your single GPU and dual GPU performance is most likely
due to ECC. This looks to me like you have ECC turned on on the nodes. I
suggest turning this OFF. I have NEVER seen an ECC error occur on a GPU that
wasn't faulty in the first place (overclocked engineering samples etc) and
to quote Seymour Cray "Parity is for farmers."
Run this as root on all your nodes and then reboot:
nvidia-smi -g 0 --ecc-config=0
nvidia-smi -g 1 --ecc-config=0
This should get you up to the 1 and 2 GPU numbers given above and will
likely boost your 4 GPU number to something closer to 4.0.
All the best
Ross
/\
\/
|\oss Walker
---------------------------------------------------------
| Assistant Research Professor |
| San Diego Supercomputer Center |
| Adjunct Assistant Professor |
| Dept. of Chemistry and Biochemistry |
| University of California San Diego |
| NVIDIA Fellow |
|
http://www.rosswalker.co.uk |
http://www.wmd-lab.org/ |
| Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
---------------------------------------------------------
Note: Electronic Mail is not secure, has no guarantee of delivery, may not
be read every day, and should not be used for urgent or sensitive issues.
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Sep 16 2011 - 10:00:03 PDT
