I was talking about changing the work fetch policy. The CPU scheduling
policy should be unaffected by this change. Work fetch can under a couple
of circumstances that appear to occur on one of my machines every few
weeks, cause not enough work to be downloaded.
jm7
David Anderson
<[email protected]
ey.edu> To
Sent by: [email protected]
<boinc_dev-bounce cc
[email protected] [email protected],
u> [email protected]
Subject
Re: [boinc_dev] Worst case for run
09/30/2010 11:28 order and work fetch.
PM
I don't understand the following.
Are you talking about changing the simulation,
or changing the scheduling policy?
The goal of the simulation is to predict what the scheduler will do.
It does this fine when there are no EDF jobs.
On 30-Sep-2010 4:34 PM, [email protected] wrote:
> It can also happen if the Short Term Debts are such that the long job
gets
> run late in the disconnected period. I don't believe that Round Robin is
> the correct method at all. A much simpler solution is to just sort the
> tasks by estimated duration into CPU buckets. This is much simpler than
> the round robin simulation.
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.
