On Sat, Mar 6, 2010 at 12:26 PM, Ross Walker <ross.rosswalker.co.uk> wrote:
> Hi Jason,
>
> You need to use the 3.0 toolkit which NVIDIA still hasn't put on their
> website. I'll send you a release candidate version shortly and you can try
> that. Can you also try the test cases. Right now all the PME ones will be
> broken due to the fact the orders of summation have changed drastically but
> all the GB test cases should pass.
I was just getting on here to respond that one of our tesla nodes has
the 3.0b toolkit installed on it, and it worked just fine there, but
it appears that you beat me to it. I knew this was probably a
possibility, but I had used pmemd.cuda with the 2.3 toolkit a couple
months ago (i version i checked out on 1/27 worked just fine).
However, my GB simulation is working just fine with the 3.0 toolkit.
This is available on nvidia's website, but only as a beta download as
far as I knew (i know i've downloaded 3.0 toolkit).
A lot of the tests passed with the 2.3 toolkit (some had the NaN error
everywhere and was just awful). With the 3.0 toolkit, the only
failures were precision differences. They were slightly more
pronounced than the errors in regular sander, but I'll chalk that up
to the single precision being used for the operations. They were
still only differing in the hundredths/thousandths place for values
that were as large as 30,000.
Perhaps I didn't allow the tests to go all the way through the PME
ones... I'll do a more thorough test on Tuesday. A note on compiling,
however. ./configure linux_em64t_cuda gfortran nopar did not work.
It failed at the link step, making some complaint about a reference to
MAIN_ or something. I don't quite recall... It was resolved when I
used the linux_em64t_cuda_SPDP.gfortran config data. If you want/need
more information, I'll be happy to try and re-create the error.
Thanks!
Jason
>
> All the best
> Ross
>
>> -----Original Message-----
>> From: amber-developers-bounces.ambermd.org [mailto:amber-developers-
>> bounces.ambermd.org] On Behalf Of Jason Swails
>> Sent: Saturday, March 06, 2010 7:54 AM
>> To: AMBER Developers Mailing List
>> Subject: [AMBER-Developers] pmemd.cuda errors
>>
>> Hello,
>>
>> I'm getting some pretty strange behavior from pmemd.cuda. I'm
>> restarting from a restrt file created by a pmemd.cuda that a month or
>> two ago, and when I do, every statement printed to the mdout looks
>> like this:
>>
>> NSTEP = 1000 TIME(PS) = 18602.000 TEMP(K) = NaN PRESS =
>> 0.0
>> Etot = NaN EKtot = NaN EPtot =
>> 279370.1445
>> BOND = 0.0000 ANGLE = 309468.3554 DIHED =
>> 0.0000
>> 1-4 NB = 0.0000 1-4 EEL = 0.0000 VDWAALS =
>> 0.0000
>> EELEC = 0.0000 EGB = -30098.2109 RESTRAINT =
>> 0.0000
>> ----------------------------------------------------------------------
>> --------
>>
>> Also, the restart of that simulation has "NaN" for every coordinate of
>> every atom. Is this expected behavior? The tesla I'm using has cuda
>> 2.3 installed.
>>
>> Thanks!
>> Jason
>>
>> --
>> ---------------------------------------
>> Jason M. Swails
>> Quantum Theory Project,
>> University of Florida
>> Ph.D. Graduate Student
>> 352-392-4032
>>
>> _______________________________________________
>> AMBER-Developers mailing list
>> AMBER-Developers.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
--
---------------------------------------
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Sat Mar 06 2010 - 10:00:03 PST
