Skip to content

Home > Science > Benchmark

AMBER 6 / PMEMD

Benchmark

Analyse de performances

Thursday 7 August 2003

Voici les performances obtenues sur le cluster Beowulf de l’UMR 8601 (équipe Kozelka).



Relative performance analysis of sander6 vs pmemd

System : DHFR, also known as JAC but with cutoff=8.0 instead of 9.0 and skinnb=1.0 instead of 2, as chosen in PMEMD benchmarks.

23558 atoms - 7182 molecules - Box : 64x64x64 Ang.

1000 steps of dynamics run - time is in ps/day.


Note that this benchmark uses a 1 fs timestep therefore the calculation is for 1 ns of trajectory.


Processor(s) Clock SANDER 6 PMEMD* PMEMD PMEMD_P4 Imp.
1 athlon 1.2Ghz 47.0 (0.57 x) -n. d. - -n. d. - -n. d. - -n. d. -
2 athlons 1.2Ghz 82.0 (1.00 x) 154.0 (1.00 x) 154.8 (1.00 x) -n. d. - 1.88
4 athlons 1.2Ghz 140.0 (1.71 x) 269.2 (1.75 x) 263.4 (1.70 x) -n. d. - 1.92
6 athlons 1.2Ghz 173.1 (2.11 x) 360.0 (2.34 x) 347.0 (2.24 x) -n. d. - 2.08
8 athlons 1.2Ghz 218.7 (2.67 x) 452.4 (2.94 x) 432.0 (2.79 x) -n. d. - 2.07
Processor(s) Clock SANDER 6 PMEMD* PMEMD PMEMD_P4 Imp.
1 xeon 2.8Ghz 78.5 (0.60 x) -n. d. - -n. d. - -n. d. - -n. d. -
2 xeons 2.8Ghz 130.5 (1.00 x) 245.5 (1.00 x) 249.0 (1.00 x) 261.8 (1.00 x) 2.01
4 xeons 2.8Ghz 213.9 (1.64 x) 413.4 (1.68 x) 417.4 (1.68 x) 440.8 (1.68 x) 2.06
6 xeons 2.8Ghz 265.8 (2.04 x) 572.2 (2.33 x) 564.7 (2.27 x) 600.0 (2.29 x) 2.26
Processor(s) Clock SANDER 6 PMEMD* PMEMD PMEMD_P4 Imp.
14 processors 317.6 744.8 822.9 -n. d. - 2.59
perf. vs 2 athlons : (3.87 x) (4.84 x) (5.32 x) -n. d. -
perf. vs 2 xeons : (2.43 x) (3.03 x) (3.30 x) -n. d. -

PMEMD* indicates PMED has been compiled with the
option -DSLOW_NONBLOCKING_MPI


Improvement of performances between sander 6 and pmemd is taken from the best pmemd time, for xeon, it is pmemd compiled specifically with pentium 4 optimisations,
for all the nodes, PMEMD without the option -DSLOW_NONBLOCKING_MPI.

Each run has been run independently -alone on the whole cluster-.



Attached documents

Comment on this article