Re: [AMBER] No hybrid REMD for cuda?

From: Hugo Azurmendi <>
Date: Wed, 3 Sep 2014 14:07:14 -0700

too bad about hybrid remd on cuda.

I'm doing small polysaccharides, the T generator is for proteins and couldn't get good results. I just found out that I may be having issues with the starting structures for some Ts. I think I'll be able to work it out.

On Wednesday, September 3, 2014 4:30 PM, Jason Swails <> wrote:

On Wed, Sep 3, 2014 at 3:53 PM, Hugo Azurmendi <> wrote:

> I'm trying to run a hybrid T-replica exchange on Amber 14 using
> pmemd.cuda.MPI in a GPU, but the parameters numwatkeep and hybridgb
> apparently are not recognized.

Hybrid REMD is supported only in sander.  It will require a lot of work to
get it working on GPUs.

> ​​
> I didn't find information on this on the manual nor a cuda test in
> $AMBERHOME. My system is relatively small and the T_remd runs OK
> the
> ​​
> mentioned parameters) , but I'm having problems to set temp0 to have
> relatively uniform exchanges through the replicas (trying with 10 now) and
> ​​
> ​​
> I believe using a hybrid EP strategy might help to improve exchange rates.

​Have you used the T-REMD temperature generator (there is a link on the
tutorial page).  In my experience, that gets really close to the desired
exchange rates (like, within 2-5%).


Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
AMBER mailing list
Received on Wed Sep 03 2014 - 14:30:02 PDT
Custom Search