[AMBER] use QUICK for QM/MM simulation accelerated by gpu

From: wangjianing_mail <wangjianing_mail.163.com>
Date: Thu, 17 Jun 2021 18:37:40 +0800 (GMT+08:00)

Dear Amber user:

When performing H-REMD sampling and using the CPU across nodes, it is found that the calculation efficiency of different sampling windows/trajectories is significantly different, the fastest is 0.21ns/d, and the slowest is 0.07ns/d.

Is the copy exchange of H-REMD in accordance with SAMPLING TIME (for example: 1ps) or in accordance with a certain REAL TIME?

If it is according to SAMPLING TIME, when the first trajectory runs to the SAMPLING TIME that will be exchanged, will this fastest trajectory wait for the slower trajectory?

Jia-Ning Wang

AMBER mailing list
Received on Thu Jun 17 2021 - 04:00:02 PDT
Custom Search