While the APR's for other projects applications are not directly comparable, the relative values ARE used by BOINC in the REC (recent estimated credit) calculations that are used in determining how much work is requested from each project relative to the resource share.

This can have a direct impact in overcommitment of the host. BTW, also beneficial is reducing the default 10 day averaging of REC in BOINC's cc_config.xml file to 1 day. Makes work requests more balanced and responsive to crunching on the host among several projects run concurrently on a host.

If the large cache of tasks are from N-Body, then I agree you are unlikely to finish most of them before deadline.

---

If the large cache of tasks are from N-Body, then I agree you are unlikely to finish most of them before deadline.

Here is the distribution of Milky Way tasks:
Application: All (322) · Milkyway@home N-Body Simulation (201) · Milkyway@home Separation (121)

Some tasks run for several hours with the elapsed time increasing, the remaining time also increasing (going backwards), with progress showing 2% or 3% finished, but no progress being made for maybe a half-hour.

I have seen that such a task will take multiple days -- as much as 17 days-- to finish. I don't have the time or patience to wait 7 days for a task to complete.

So far, I have aborted five of those.

I will attempt to follow your suggestion to reduce "...the default 10 day averaging of REC in BOINC's cc_config.xml file to 1 day," without doing any damage.

Steven Gaber

If the large cache of tasks are from N-Body, then I agree you are unlikely to finish most of them before deadline.

Here is the distribution of Milky Way tasks:
Application: All (322) · Milkyway@home N-Body Simulation (201) · Milkyway@home Separation (121)

Some tasks run for several hours with the elapsed time increasing, the remaining time also increasing (going backwards), with progress showing 2% or 3% finished, but no progress being made for maybe a half-hour.

I have seen that such a task will take multiple days -- as much as 17 days-- to finish. I don't have the time or patience to wait 7 days for a task to complete.

So far, I have aborted five of those.

I will attempt to follow your suggestion to reduce "...the default 10 day averaging of REC in BOINC's cc_config.xml file to 1 day," without doing any damage.

Steven Gaber

If you just change the cache size MilkyWay will stop sending you new tasks until your new cache size is reached.

[/quote] If you just change the cache size MilkyWay will stop sending you new tasks until your new cache size is reached.[/quote]

Yeah, I did as Keith suggested, reduce the cache size down to 0.01 days and 0.0 additional days. But it was too late, Milky Way sent me 329 tasks, all due by March 10.

After it sent me that huge slug of tasks, I specified No New Tasks until it whittles down the load. The rest will just time out.

I could abort most 201 of the N-Body ones.

Just now I increased the usage limit to 100% and 90% of the CPUs as an experiment. I'll see what happens next.

You can tell I really don't know what I'm doing. I just follow you guys' suggestions.

Steven Gader