[Beowulf] Please help to setup Beowulf
reuti at staff.uni-marburg.de
Wed Feb 18 01:42:19 PST 2009
Am 18.02.2009 um 07:32 schrieb Mark Hahn:
>> searches. Without array task scheduling this would require 500,000
>> individual job submissions. The fact that I never met a serious
>> PBS shop that had not
> what's wrong with 500k job submissions? to me, the existence of
> "array jobs"
> is an admission that the job/queueing system is inefficient.
When I compare this e.g. to C:
- a loop like:
printf("Hello from run %d.\n", i);
- and you can guess: 100,000 times:
printf("Hello from run %d.\n", i++);
While the execution time is nearly the same, compilation of the first
one is faster by far. So, that a "for"-instruction in C exists is
also an admission that it can't compile sequential code very well and
generates a big executable? The compiler must read the source, and
SGE has to read the job requirements for every job again and again
and store it.
(PS: not to mention, that a for-loop/array-job is easier to handle
for the user)
> if you're saying that the issue is not per-job overhead of
> submission, but rather that jobs are too short, well, I think
> that's a user problem. I think it's entirely reasonable to require
> user jobs to consume some minimum cpu time
> (say, few minutes).
>> - Policy and resource allocation features are very important to
>> people deploying these systems
> so I'm curious what that means. things like "dept A needs to be
> N cpus, but dept B gets to use whatever is left over"? or node
> choice based on amount of free disk? I don't really see why these
> sorts of issues
> would be less important to more parallel environments.
>> - Storage speed is often more important than network speed or
>> latency in many cases
> which makes me wonder: do bio types consider using map-reduce-like
> frameworks? that is, basically distributing the work to the data.
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit
More information about the Beowulf