[AMBER] MD GPU Scaling Performance

From: Jatin Kashyap <jk435.njit.edu>
Date: Thu, 7 Jan 2021 17:38:26 -0500

Dear AMBER Community,

I am conducting GPU scaling studies on a system containing around 50k atoms with protein-ligand/water system.
My admin suggested that from their past experience, the highest performance has been achieved by using single GPU node with maximum GPU cores on it i.e. 4, on their machine.

And by following it, I performed study on 1,2,3,4 cores on single node(hitting maximum node capacity). Yes I do get the fastest performance on 4 GPU but still it doesn’t look like fast enough.
It is giving 1.3 ns/day and with that speed it will takes months for me to perform 200 ns trajectory. Is it normal behavior or am I doing something very wrong i.e. shall go above 4 GPU cores etc…

Thank you for the guidance.

Jatin Kashyap
Ph.D. Student
Dr. Dibakar Datta Group
Department of Mechanical and Industrial Engineering
New Jersey Institute of Technology (NJIT)
University Heights
Newark, NJ 07102-1982
Phone- (201)889-5783
Email- jk435.njit.edu

AMBER mailing list
Received on Thu Jan 07 2021 - 15:00:02 PST
Custom Search