Dear All,
For those who are interested our provisional tests (Cellulose NPT,
FactorIX NVE) are NOT showing problems on GTX780s like they do on the
GTX-Titan cards. We are still testing things, including multi-GPU so this
is only provisional at this point. We will provide an update once we have
conducted more in-depth testing but for now at least things are looking
positive.
The following are some provisional benchmark numbers for AMBER 12 (GPU
v12.3 - bug fix 18) on the 4 x GTX780 Exxact Test Drive Machine - See the
following page for specs on this machine
(
http://ambermd.org/gpus/recommended_hardware.htm#exxact_machines) which
is the same as the GTX680 workstation but with GTX780 cards.
JAC_PRODUCTION_NVE - 23,558 atoms PME
-------------------------------------
CPU code 16 cores: | ns/day = 13.73 seconds/ns = 6293.07
1 x GTX780: | ns/day = 98.96 seconds/ns = 873.12
2 x GTX780: | ns/day = 105.21 seconds/ns = 821.18
3 x GTX780: | ns/day = 116.51 seconds/ns = 741.55
4 x GTX780: | ns/day = 135.81 seconds/ns = 636.16
JAC_PRODUCTION_NPT - 23,558 atoms PME
-------------------------------------
CPU code 16 cores: | ns/day = 11.53 seconds/ns = 7491.94
1 x GTX780: | ns/day = 77.11 seconds/ns = 1120.45
2 x GTX780: | ns/day = 83.18 seconds/ns = 1038.74
3 x GTX780: | ns/day = 98.84 seconds/ns = 874.12
4 x GTX780: | ns/day = 110.95 seconds/ns = 778.76
FACTOR_IX_PRODUCTION_NVE - 90,906 atoms PME
-------------------------------------------
CPU code 16 cores: | ns/day = 3.81 seconds/ns = 22670.04
1 x GTX780: | ns/day = 28.82 seconds/ns = 2998.01
2 x GTX780: | ns/day = 33.42 seconds/ns = 2585.47
3 x GTX780: | ns/day = 38.67 seconds/ns = 2234.55
4 x GTX780: | ns/day = 44.32 seconds/ns = 1949.54
FACTOR_IX_PRODUCTION_NPT - 90,906 atoms PME
-------------------------------------------
CPU code 16 cores: | ns/day = 3.18 seconds/ns = 27175.15
1 x GTX780: | ns/day = 23.22 seconds/ns = 3720.92
2 x GTX780: | ns/day = 28.15 seconds/ns = 3068.77
3 x GTX780: | ns/day = 31.58 seconds/ns = 2735.63
4 x GTX780: | ns/day = 38.26 seconds/ns = 2258.34
CELLULOSE_PRODUCTION_NVE - 408,609 atoms PME
--------------------------------------------
CPU code 16 cores: | ns/day = 0.86 seconds/ns =
100654.62
1 x GTX780: | ns/day = 6.47 seconds/ns =
13360.72
2 x GTX780: | ns/day = 7.34 seconds/ns =
11763.50
3 x GTX780: | ns/day = 8.50 seconds/ns =
10164.60
4 x GTX780: | ns/day = 9.57 seconds/ns = 9027.83
CELLULOSE_PRODUCTION_NPT - 408,609 atoms PME
--------------------------------------------
CPU code 16 cores: | ns/day = 0.69 seconds/ns =
125740.42
1 x GTX780: | ns/day = 5.34 seconds/ns = 16177.63
2 x GTX780: | ns/day = 6.24 seconds/ns =
13843.73
3 x GTX780: | ns/day = 7.37 seconds/ns =
11726.06
4 x GTX780: | ns/day = 8.25 seconds/ns =
10477.60
TRPCAGE_PRODUCTION - 304 atoms GB
---------------------------------
CPU code 16 cores: | ns/day = 168.55 seconds/ns = 512.62
1 x GTX780: | ns/day = 676.35 seconds/ns = 127.74
2 x GTX780: N/A
3 x GTX780: N/A
4 x GTX780: N/A
MYOGLOBIN_PRODUCTION - 2,492 atoms GB
-------------------------------------
CPU code 16 cores: | ns/day = 5.50 seconds/ns =
15718.91
1 x GTX780: | ns/day = 175.66 seconds/ns = 491.87
2 x GTX780: | ns/day = 183.86 seconds/ns = 469.93
3 x GTX780: | ns/day = 199.55 seconds/ns = 432.98
4 x GTX780: | ns/day = 188.26 seconds/ns = 458.95
NUCLEOSOME_PRODUCTION - 25,095 atoms GB
---------------------------------------
CPU code 16 cores: | ns/day = 0.06 seconds/ns =
1417049.08
1 x GTX780: | ns/day = 3.19 seconds/ns = 27126.35
2 x GTX780: | ns/day = 4.11 seconds/ns = 21004.39
3 x GTX780: | ns/day = 5.33 seconds/ns = 16214.22
4 x GTX780: | ns/day = 6.72 seconds/ns = 12857.94
All the best
Ross
/\
\/
|\oss Walker
---------------------------------------------------------
| Associate Research Professor |
| San Diego Supercomputer Center |
| Adjunct Associate 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 Mon Jun 24 2013 - 14:30:02 PDT
