Re: [Condor-users] GPU and condor?

[Xiang Ni <nixiang.nn@xxxxxxxxx>]

Hello Tung-Han Hsieh,

Very interesting! You just change the configure file and everything
works. It seems that there is some work to do to track the status of
the GPU. I'm wondering how do you do that in your system. By adding
codes to ask GPU to send heartbeat information to Condor?

Thanks!

2010/1/7 Miron Livny <miron@xxxxxxxxxxx>:
> To all GPUers out there,
>
> We would be very interested in hearing from you what Condor can do to help
> you in managing GPU clusters. So far we did not find much we can offer in
> this space. Any guidance you can provide will be most welcomed.
>
> Miron
>
>
>
> Xiang Ni wrote:
>>
>> Hi Tung-Han Hsieh,
>>
>> Thanks and you sharing is very helpful!
>>
>> I'm also interested in this topic and I have some confusions.
>>
>> How do you make condor aware of the existence of GPUs? By modifying the
>> Hawkeye?
>>
>> Thanks!
>>
>> Regards,
>>
>> 2010/1/7 Tung-Han Hsieh <tunghan.hsieh@xxxxxxxxx>:
>>>
>>> Hello,
>>>
>>> We have some experiences on building a GPU cluster using
>>> condor.
>>>
>>> Currently we have two GPU clusters, used for different
>>> research
>>> groups. Each cluster is composed by the following
>>> element:
>>>
>>> 1. Head node: Running condor server, for users login to
>>> build
>>>              their codes, submit jobs,
>>> etc.
>>>
>>> 2. File servers: The Lustre Cluster filesystems are
>>> deployeed.
>>>
>>> 3. Computing nodes: Each node has at least one, at most 4
>>> GPUs.
>>>                    Each cluster has more than 64 GPUs
>>> installed.
>>>
>>> 4. Communication: one has infiniband network, and the other
>>> use
>>>                  Gigabit
>>> network.
>>>
>>> The condor system can allocate multi-GPUs for users. In
>>> our
>>> implimentation the number of CPU cores in each computing
>>> node
>>> is not important. So in condor command file, users
>>> specify
>>> "machine_count" is actually specify the number of GPUs
>>> required.
>>> And the number of GPUs in each node is hard coded as
>>> the
>>> "NUM_CPUS" in the local condor config. file in each
>>> node.
>>>
>>> Honestly, we are not the condor experts. Hence we also
>>> developed
>>> some codes to help condor to do more complicated tasks, such
>>> as
>>> user quota for number of GPUs, GPU assignment, dead job cleaning,
>>> etc.
>>> But I guess all of these could be done by condor itself. We
>>> just
>>> don't know how to do, so try the somewhat stupid way to write
>>> codes
>>> to do
>>> those.
>>>
>>> Probably we can communicate the experience about this subject
>>> :)
>>>
>>>
>>> Cheers,
>>>
>>> T.H.Hsieh
>>>
>>>
>>> 2010/1/7 Marian Zvada <zvada@xxxxxxxx>
>>>>
>>>> Dear Condor Folks,
>>>>
>>>> is there someone in Condor user's community who has build GPU cluster
>>>> based on condor?
>>>> I mean someone, who has worker nodes hw with GPU graphical cards and job
>>>> management is done by condor on the top.
>>>>
>>>> We are very interested in this topic and would like to build such a
>>>> infrastructure (condor + gpu worker nodes) for research people in our
>>>> organization.
>>>> In first epoch of this project we'd like to develop standalone cluster:
>>>>
>>>> - master condor head node
>>>> - 5 gpu worker nodes (each worker node 2x nVIDIA GTX295)
>>>> - storage element for data
>>>>
>>>> I know, there is a lot to see on google about such a experiments, but I
>>>> wanted to ask directly from condor users about their
>>>> opinions/suggestions/recommendations since we are serious about to build
>>>> condor gpu cluster and use it in production for our research activities.
>>>>
>>>> If there is someone who has done similar setup and is willing share the
>>>> knowledge, I appreciate talk about it! Any url hints are welcome too...
>>>>
>>>> Thanks and regards,
>>>> Marian
>>>> _______________________________________________
>>>> Condor-users mailing list
>>>> To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx with
>>>> a
>>>> subject: Unsubscribe
>>>> You can also unsubscribe by visiting
>>>> https://lists.cs.wisc.edu/mailman/listinfo/condor-users
>>>>
>>>> The archives can be found at:
>>>> https://lists.cs.wisc.edu/archive/condor-users/
>>>
>>> _______________________________________________
>>> Condor-users mailing list
>>> To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx with a
>>> subject: Unsubscribe
>>> You can also unsubscribe by visiting
>>> https://lists.cs.wisc.edu/mailman/listinfo/condor-users
>>>
>>> The archives can be found at:
>>> https://lists.cs.wisc.edu/archive/condor-users/
>>>
>>>
>>
>>
>>
> _______________________________________________
> Condor-users mailing list
> To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx with a
> subject: Unsubscribe
> You can also unsubscribe by visiting
> https://lists.cs.wisc.edu/mailman/listinfo/condor-users
>
> The archives can be found at:
> https://lists.cs.wisc.edu/archive/condor-users/
>



-- 
Xiang Ni
Sino-German Joint Software Institute
Computer Science&Engineer Deparment of Beihang University
100191