Dynamic switching GPU/CPU minimization

From: Maximilian Ebert (mebert_at_chemcomp.com)
Date: Sat Jul 11 2020 - 08:49:50 CDT

Next message: Raman Preet Singh: "Restarting aborted NAMD run"
Previous message: Alex Balaeff: "Pair interaction calculations of the QM/MM system"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

Dear list,

I recently had some minimizations and MD runs failed during testing of our new GPUs. After minimization on the CPU and disabled GPU the protocol worked fine. I always thought that NAMD would use the GPU in single precision mode during energy minimization only if the energy difference between two steps falls below a certain value. Now I am not sure anymore. Could anyone explain how NAMD handles large clashes when using energy minimization on a GPU? Is there a dynamic switching between CPU/GPU? Is there a way activate double precision mode for the energy minimization?

Thanks,
Max

------
Maximilian Ebert
Application Scientist
Chemical Computing Group | www.chemcomp.com mebert_at_chemcomp.com | +1 (514) 393-1055 For support inquiries: support_at_chemcomp.com

------
Chemical Computing Group | www.chemcomp.com |
910-1010 Sherbrooke W., Montreal QC H3A 2R7, Canada | +1 514 393 1055

Manage Communications: www.chemcomp.com/Manage_Subscription.htm
t

-- 
Chemical Computing Group | www.chemcomp.com <http://www.chemcomp.com> | 
910-1010 Sherbrooke W., Montreal QC H3A 2R7, Canada | +1 514 393 1055
Manage Communications: www.chemcomp.com/Manage_Subscription.htm 
<http://www.chemcomp.com/Manage_Subscription.htm>

Next message: Raman Preet Singh: "Restarting aborted NAMD run"
Previous message: Alex Balaeff: "Pair interaction calculations of the QM/MM system"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]

This archive was generated by hypermail 2.1.6 : Fri Dec 31 2021 - 23:17:09 CST