Hi Ross and Scott,
Thank you for the details and benchmarks! Really great performance improvement!! Hope that will be able to use the new code soon!
The only think that is a bit confusing for me is the scale. According to the benchmark the scale between one and 2xM2090 is just 19%, but between 2xM2090 and 4xM2090 is 50%, what could be the reason? Actually from these numbers the scale for all systems is a bit worse compared to the previous benchmark results. Is that due to the new hardware (the new M2090 cards) or due to the new code changes?
Regards,
Filip
--- On Thu, 5/19/11, Ross Walker <ross.rosswalker.co.uk> wrote:
> From: Ross Walker <ross.rosswalker.co.uk>
> Subject: Re: [AMBER] Amber 11 Benchmark for GPUs - Binaries and an Evaluation Licenses
> To: "'AMBER Mailing List'" <amber.ambermd.org>
> Date: Thursday, May 19, 2011, 7:54 PM
> Hi Filip,
>
> > > to the AMBER 11 GPU support in a few weeks that
> massively
> > > boosts performance
> >
> > Great news!! Could tell us/reveal a bit more about
> this new update?
> > Some preliminary details and what we could expect in
> terms of the boost
> > performance?
>
> This will be a big performance improvement patch (about 1.9
> to 2.1x for all
> system sizes [PME]) as well as addressing a few minor bugs.
> Additional
> features including REMD, Constant-PH, NMR Restraints,
> Umbrella Sampling and
> Extra point support will hopefully be part of AMBER 12 if
> Adrian and I are
> successful in getting our NSF-SI2-SSE grants renewed.
>
> Here are 'provisional' numbers for the updated code
> (ns/day):
>
>
> AMBER 11
> AMBER 11 Update
> JAC NVE
> (1xC2050) 21.29
> 37.91
>
> (1xM2090)
> 43.57
>
> (2xM2090)
> 58.50
>
> (4xM2090)
> 78.06
>
> (8xM2090)
> 88.71
>
> FactorIX NVE (1xC2050) 5.73
> 11.19
>
> (1xM2090)
> 13.29
>
> (2xM2090)
> 15.79
>
> (4xM2090)
> 23.64
>
> We are just testing the updates to make sure they are ready
> for release and
> then I'll update the webpage with new benchmarks.
>
> 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
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu May 19 2011 - 13:30:04 PDT
