AW: NAMD Multicore CUDA usage

From: Norman Geist (norman.geist_at_uni-greifswald.de)
Date: Tue Jun 10 2014 - 02:33:13 CDT

Try:

 

fullelectfrequency 4

 

which may require adding a slight temperature control:

 

langevin on

langevintemp 300

langevindamping 0.01

 

Now:

 

namd2 +idlepoll +p64 +devices 0,1,2,3 +ignoresharing eq02.conf >& eq02.log

 

Depending on system size try within namd config file:

 

twoawayx yes

 

If that brings improvement, also try:

 

twoawayx yes

twoawayy yes

 

Now play along with the number of cores and gpus and see what works best.

 

Norman Geist.

 

Von: owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] Im Auftrag von Ravi Abrol
Gesendet: Freitag, 6. Juni 2014 21:20
An: namd-l_at_ks.uiuc.edu
Betreff: namd-l: NAMD Multicore CUDA usage

 

Dear Users,

I have a 64-core node with 4 GPUs and I am running MD on an 80K system.
Is there a good combination of CPUs+GPUs that gives best performance in terms
of ns/day? I want to learn from your experience.

With the following command:
namd2 +idlepoll +p16 +devices 0 eq02.conf >& eq02.log

I could get 0.4days/ns and I don't know if I can improve on it, as adding more GPUs or CPUs
didn't help.
Also, I am not sure if I am missing any command line options that might

help with a performance boost.

Thanks,

Ravi

 

---
Diese E-Mail ist frei von Viren und Malware, denn der avast! Antivirus Schutz ist aktiv.
http://www.avast.com

This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:22:27 CST