30 reputation
8
bio website N/A
location Germany
age 27
visits member for 3 years, 11 months
seen Jul 31 at 10:28

What's there to say about me? I'm a computer science expert from Germany with a broad range of interests when it comes to information technology... I consider myself a geek but not a nerd and a hacker but not a cracker and apart from computer stuff music may be my number one hobby.


Feb
14
awarded  Popular Question
Dec
7
awarded  Popular Question
Feb
1
comment OpenMPI in SGE fails when not observed
here you are... nothing fancy, really
Feb
1
revised OpenMPI in SGE fails when not observed
added 536 characters in body
Feb
1
answered OpenMPI in SGE fails when not observed
Jan
31
asked OpenMPI in SGE fails when not observed
Nov
22
comment Howto set up SGE for CUDA devices?
pastebin.com/k35KNeyT - that's an updated version of the script using python and the xml output of nvidia-smi... I guess the xml isn't as likely to change as the pretty print standard output
Oct
19
comment Howto set up SGE for CUDA devices?
I've already edited my original question to share that solution... the essence of it is: it only works if you put your GPUs in EXCLUSIVE_THREAD mode... then you'll get the number of used GPUs with the following command: nvidia-smi |grep "Process name" -A 9|grep -v +-|grep -v \|=|grep -v Usage|grep -v "No running"|wc -l
Oct
18
awarded  Editor
Oct
18
revised Howto set up SGE for CUDA devices?
added information on how to set up load monitoring for the GPUs
Oct
18
comment Howto set up SGE for CUDA devices?
nevermind... I've been able to create a load sensor for this :)
Oct
18
accepted Howto set up SGE for CUDA devices?
Oct
18
comment Howto set up SGE for CUDA devices?
Thank you! That definitely helped me get it working so I gladly accept your answer... There's one downside to this solution though... it depends on what the users say they would use. If you can point me to some elegant solution where the gpu resource is actually monitored and the -l parameter is only used to schedule the job instead of also being used to "calculate" how many GPUs are left, that would be great :)
Oct
17
asked Howto set up SGE for CUDA devices?
Oct
17
comment Parallel Environment (PE) on Sun Grid Engine (6.2u5) won't run jobs: “only offers 0 slots”
is your "A" server also defined as execution host?
Jun
7
awarded  Scholar
Jun
7
accepted Windows DHCP-Client shall not send UID
Jun
7
comment Windows DHCP-Client shall not send UID
i also found this but it's 5 years old information... there may have been the remote chance that this situation had changed in the meantime...
Jun
7
comment Windows DHCP-Client shall not send UID
i know... that's what i already said... all i want is to prevent windows from using a uid at all... btw: pxe doesn't use uids either (even if you'd like to) and pxe is younger than rfc 2131
Jun
7
comment Windows DHCP-Client shall not send UID
linux let's you, too... but if it's possible to stop windows from using it, it's less bothersome than to set a unique (per machine) one for windows and linux. it's one additional step after restoring an image that could be avoided if it was possible to stop windows from using it