Re: [AMBER] CUDA (GPU): Implementation does not support extra points.

From: Ross Walker <ross.rosswalker.co.uk>
Date: Sat, 12 Jun 2010 11:42:30 -0700

Hi Ashutosh

Granted AMBER's error messages can be confusing and unclear at times but
surely this is about as clear cut an error message as you can hope for.

You are using extra points (TIP4P/5P or other extra points force field) and
this is NOT supported by the GPU implementation. Build your system WITHOUT
extra points (solvate with TIP3P for example) or use the regular MPI code.

A complete list of limitations is here:

http://ambermd.org/gpus/

All the best
Ross

> -----Original Message-----
> From: ashutosh shandilya [mailto:izerokelvin.gmail.com]
> Sent: Saturday, June 12, 2010 11:18 AM
> To: AMBER Mailing List
> Subject: [AMBER] CUDA (GPU): Implementation does not support extra
> points.
>
> Hi all
> I am running pmemd on amber 11 with cuda it gives the error
>
> CUDA (GPU): Implementation does not support extra points.
> Require numextra == 0.
>
> Here is the input file:
>
> 2.5ps MD equilibration on protein
> &cntrl
> imin = 0,
> irest = 0,
> ntx=1,
> ntb=1,
> cut = 10,
> tempi = 0.0,
> temp0 = 5.0,
> ntt = 3,
> nstlim = 2500, dt = 0.001,
> ntpr = 2500, ntwx = 2500, ntwr = 2500
> /
>
>
> what does that mean.
> Thanks in anticipation
> regards
> Ashutosh Shandilya
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Jun 12 2010 - 12:00:04 PDT
Custom Search