RE: A home PC to run NAMD simulations

From: Sanjay Hari (shari_at_universalsequencing.com)
Date: Tue Oct 13 2020 - 15:03:07 CDT

You might want to look into a cloud-based deployment using AWS. You can get a spot instance of a 4-gpu machine for <$4/hr. There is conveniently a free NAMD AMI (big thank you to TCBG) with everything pre-compiled so there’s no hassle.

https://aws.amazon.com/marketplace/pp/B06VSMMHRY

From: owner-namd-l_at_ks.uiuc.edu <owner-namd-l_at_ks.uiuc.edu> On Behalf Of Adupa Vasista
Sent: Sunday, October 11, 2020 5:13 AM
To: NAMD list <namd-l_at_ks.uiuc.edu>; Harish Srinivasan <harishlglk1992_at_gmail.com>
Subject: Re: namd-l: A home PC to run NAMD simulations

Have a look at these benchmark results : https://www.pugetsystems.com/labs/hpc/RTX3080-TensorFlow-and-NAMD-Performance-on-Linux-Preliminary-1885/
https://www.pugetsystems.com/labs/hpc/AMD-Ryzen-3950x-Compute-Performance-Linpack-and-NAMD-1617/

I personally prefer AMD Ryzen processors with RTX 3080

On Sun, Oct 11, 2020 at 12:35 PM Harish Srinivasan <harishlglk1992_at_gmail.com<mailto:harishlglk1992_at_gmail.com>> wrote:
I am planning to set up a home PC to run NAMD simulations of bilyaer, bilayer-protein and ionic liquid systems. I have decided to go with the AMD Ryzen processor (for budget reasons, although I know intel would be better suited). I would like to know the kind of GPU I should go for, particularly, what role does the GPU memory play and how much of a difference does it make to choose 2080 over 1660/1650 NVIDIA cards.

 I was suggested that NVIDIA Quadro GPUs are the best suited for CUDA based calculations, any inputs about the compatibility of these cards with Ryzen processors would also be helpful.

PS: I am looking for a low budget setup since I will be using it at home!




--

This archive was generated by hypermail 2.1.6 : Fri Dec 31 2021 - 23:17:10 CST