[Beowulf] Reusing cores was Re: Bright Cluster Manager

[Lux, Jim (337K)]

The best successes I have seen on clusters is where the heavy parallel
applications get exclusive compute nodes. Cleaner, you get all the memory
and storage bandwidth and easy to clean up. Hell, reboot the things after
each job. You got an exclusive node.

You are describing the BlueGene/Q philosophy there John. :-)

This idea tends to break when you throw GPUs in to the mix as there
(hopefully) you only need a couple of cores on the node to shovel data around
and the GPU does the gruntwork. That means you'll generally have cores left
over that could be doing something useful.

but compare the "value" of the computational work those otherwise unused cores can do versus the "cost" of a more complex system management environment. Isn't the whole idea that "hardware is cheap, wetware is expensive"

_______________________________________________
Beowulf mailing list, ***@beowulf.org sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/ma

[Chris Samuel]

but compare the "value" of the computational work those otherwise unused
cores can do versus the "cost" of a more complex system management
environment. Isn't the whole idea that "hardware is cheap, wetware is
expensive"

Ah, but the routing is transparent to our users as the filtering into the
correct partitions in Slurm is done in our submit filter.

If they request GPUs (which they must to be able to see them as we use cgroups
to control access) then they end up in the GPU partition and if they don't
then they will end up in the ordinary partition.

If they ask for too many cores per node for non-GPU jobs then they get a
message to tell them the maximum they can request.

All the best!
Chris