Re: [AMBER] pmemd.cuda segfaults

From: <pavel.banas.upol.cz>
Date: Tue, 11 Mar 2014 15:54:23 +0100 (CET)

Hi Jason,
here is an output of "lsb_release -a" on that cluster

banas.vaire:~$ lsb_release -a



No LSB modules are available.

Distributor ID: Debian

Description:    Debian GNU/Linux 7.4 (wheezy)

Release:        7.4

Codename:       wheezy

and indeed we solved the problem by downgrading kernel. Anyway, currently it
is working without any sign of errors, bugs or non-reproducible data. So,
happyend :-)

have a nice day, Pavel

-- 
Pavel Banáš
pavel.banas.upol.cz
Department of Physical Chemistry, 
Palacky University Olomouc 
Czech Republic 
---------- Původní zpráva ----------
Od: Jason Swails <jason.swails.gmail.com>
Komu: AMBER Mailing List <amber.ambermd.org>
Datum: 8. 3. 2014 20:12:16
Předmět: Re: [AMBER] pmemd.cuda segfaults
"On Sat, Mar 8, 2014 at 1:00 PM, Ross Walker <ross.rosswalker.co.uk> wrote:
> Hi Pavel,
>
> Well that's a completely new one for me. And also VERY worrying. I've
> never used Core linux (or even heard of it until now) but this makes it
> sound like some seriously Ghetto Linux Distro. Do other programs fail on
> it or just AMBER? - Either way it would scare me to use something flakey
> like that. Is it even on the list of supported Distro's for CUDA?
>
The versions that the OP posted look like they have to be Linux kernel
versions. My Linux box is on kernel version 3.10.25 right now and it has
been rock stable with a GTX 680. If my distro only makes stable kernels
available through its package manager (which I suspect is the case), only
3.2.55 from the OP's list appears to be stable. My distro also makes
versions newer and older than 3.11.8 and 3.8.13 available, but not those
specific versions.
.pavel: what distro are you actually using? Specifically, what is the
output of "lsb_release -a"? On my machine, for instance, it says:
swails.batman ~ $ lsb_release -a
LSB Version: n/a
Distributor ID: Gentoo
Description: NAME=Gentoo
Release: n/a
Codename: n/a
It would be interesting to see which distro is actually pushing those
kernels on their users. That said, my actual suspicion is that the OP's
distro was pushing a kernel update through its updater and the CUDA driver
was not rebuilt with the new kernels. I know Ubuntu does this and is one
of the principle forces that drove me away from it several years ago.
In that case, reverting to the 'original' kernel the driver was built for
would certainly appear to 'fix' the problem but does not indicate those
kernels break CUDA support. That is my suspicion.
All the best,
Jason
-- 
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
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 Tue Mar 11 2014 - 08:00:04 PDT
Custom Search