From: Giacomo Fiorin (giacomo.fiorin_at_gmail.com)
Date: Wed Jul 20 2011 - 20:41:27 CDT
Hi Asjasja, you could send the "STOP" signal with the kill command, once you
know the process IDs, and give the "CONT" signal after the break is over.
If you don't wanna trace down the PIDs with "top" (since you want to write a
script), you could use "killall":
killall -STOP namd2
sleep xx
killall -CONT namd2
On Wed, Jul 20, 2011 at 6:10 PM, Ajasja Ljubetič
<ajasja.ljubetic_at_gmail.com>wrote:
> Dear All,
>
> Is there an easy way temporarily suspend a running NAMD simulation from
> the command line? I'm running NAMD 2.8 on Ubuntu 10.04 (64 bit).
> I'm using charmrun +local ... to run my jobs.
>
> The reason I'm asking this is because I have some nodes with two GPUs,
> where one GPU keeps overheating (they are just humble GTX 560). I tried
> increasing the airflow, but I'm at the limit of what I can do. Since this is
> only a problem in the summer I would need a temporary solution. So I would
> like to have a script along the lines (in pseudo code):
>
> every 10 seconds do
> if GPU_TEMP>99 then
> suspend namd
> sleep 10 seconds
> resume namd
>
> Actually I have to ideas how I could implement this.
> a) I could just kill namd and then sleep for some time and then figuer out
> what the latest restart file is and start from there.
> b) I could add this check inside the configuration file. So I could run for
> a small number of steps, check the GPU temp and then either run for a small
> number of steps or sleep for some time. (again in pseudo code)
>
> while i<NUMBER_OF_PARTS do
> if GPU_TEMP>99 then
> sleep 30 seconds
> else
> run 100
> inc i
>
> Any other suggestions are welcome.
>
> Best regards,
> Ajasja Ljubetic
> Young researcher
> Laboratory of biophysics
> Department F5
> Institute Jozef Stefan
> Ljubljana, Slovenia
>
This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:20:36 CST