Oh yeah - forgot to say - the people running these big machines won't
(without justification I might add) let you turn off ECC on their cards so
performance will always be about 8 to 10% lower than the numbers on the
AMBER benchmark page - which is what you will get if you have your own
machine with K20's in and configure it the 'recommended' way.
Thanks for posting these numbers here Parker.
On 2/13/14, 9:57 AM, "de Waal, Parker" <Parker.DeWaal.vai.org> wrote:
>Hi David,
>
>Here are the single gpu benchmarks using Amber version 12.21 on Stampede:
>
>JAC_PRODUCTION_NVE - 23,558 atoms PME
>-------------------------------------
> 1 x K20: | ns/day = 75.53 seconds/ns = 1143.86
>
>JAC_PRODUCTION_NPT - 23,558 atoms PME
>-------------------------------------
> 1 x K20: | ns/day = 60.32 seconds/ns = 1432.42
>
>FACTOR_IX_PRODUCTION_NVE - 90,906 atoms PME
>-------------------------------------------
> 1 x K20: | ns/day = 20.48 seconds/ns = 4219.73
>
>FACTOR_IX_PRODUCTION_NPT - 90,906 atoms PME
>------------------------------------------
> 1 x K20: | ns/day = 16.94 seconds/ns = 5099.55
>
>CELLULOSE_PRODUCTION_NVE - 408,609 atoms PME
>--------------------------------------------
> 1 x K20: | ns/day = 4.99 seconds/ns = 17308.18
>
>CELLULOSE_PRODUCTION_NPT - 408,609 atoms PME
>--------------------------------------------
> 1 x K20: | ns/day = 4.27 seconds/ns = 20229.25
>
>TRPCAGE_PRODUCTION - 304 atoms GB
>---------------------------------
> 1 x K20: | ns/day = 553.86 seconds/ns = 156.00
>
>MYOGLOBIN_PRODUCTION - 2,492 atoms GB
>-------------------------------------
> 1 x K20: | ns/day = 162.44 seconds/ns = 531.89
>
>NUCLEOSOME_PRODUCTION - 25,095 atoms GB
>---------------------------------------
> 1 x K20: | ns/day = 2.92 seconds/ns = 29584.33
>
>Hope this helps,
>Parker
>
>
>
>-----Original Message-----
>From: David Cowburn [mailto:david.cowburn.einstein.yu.edu]
>Sent: Thursday, February 13, 2014 11:41 AM
>To: amber.ambermd.org
>Subject: [AMBER] TACC Stampede AMBER
>
>Does anyone have benchmarks for AMBER-12 on Stampede at TACC?
>Anticipatory thanks, David .
>
>David Cowburn, PhD DSC, Professor, Depts. of Biochemistry and Physiology
>& Biophysics Albert Einstein College of Medicine of Yeshiva University
>1300 Morris Park Avenue, Bronx, NY 10461
>gV: 908 913 0495
>
>_______________________________________________
>AMBER mailing list
>AMBER.ambermd.org
>http://scanmail.trustwave.com/?c=129&d=mvX80gdXAOTHpRhI-Z9jOaqsf_GvA1dXa-Y
>lU82h9A&u=http%3a%2f%2flists%2eambermd%2eorg%2fmailman%2flistinfo%2famber
>
>_______________________________________________
>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 Feb 13 2014 - 10:30:02 PST
