Despite having loads of disk space and the cache set to 10 days, I can only ever download 1 days work as Boinc keeps telling me the computer is over committed. I'm crunching a unit in about 3h 45m so 10 days cache should give me about 60 units, whereas I only get about 6 or 7 at a time.

Any ideas anyone?

---

Ah, Muriel Miggins - will she ever be mine

Despite having loads of disk space and the cache set to 10 days, I can only ever download 1 days work as Boinc keeps telling me the computer is over committed. I'm crunching a unit in about 3h 45m so 10 days cache should give me about 60 units, whereas I only get about 6 or 7 at a time.

Your Computer Summary page is showing an average turnaround time of almost two days. A typical deadline being two weeks from the time of issue, well, you can do the math! AIUI the more results you return promptly, the smaller that number will become and the more work you'll be assigned. Only you can see some of the information on that page: what does it show you for running & working percentages, CPU efficiency, and RDCF?

Reducing your “connect every” setting, temporarily at least, could actually help, paradoxical as that may sound: as far as turnaround time is concerned, whether a WU is crunched in one hour or ten makes no difference if it won’t get returned for a week.

See the BOINC Wiki Work Buffer article.

---

[quote] Only you can see some of the information on that page: what does it show you for running & working percentages, CPU efficiency, and RDCF? [Quote]

running 99.979%
working allowed 4.6481%
CPU 0.99979
duration factor .343873

OK the problem appears to be with the 4% work allowed time that appears. How has that happened and how can I fix it?. Other computers I have on Boinc are up at the 90% mark.

---

Ah, Muriel Miggins - will she ever be mine

running 99.979%
working allowed 4.6481%
CPU 0.99979
duration factor .343873

OK the problem appears to be with the 4% work allowed time that appears. How has that happened and how can I fix it?. Other computers I have on Boinc are up at the 90% mark.

If the system’s not being that heavily used (which doesn’t seem likely), something must be making BOINC think it is. I'm not a Windows person, so I don’t have much advice to offer, but you could try hitting Control-Alt-Delete to see what processes running in the background might be interrupting BOINC. Is your “work while computer is in use” preference set to “yes”? BOINC Manager set to “run according to preferences”?

---

Is your “work while computer is in use” preference set to “yes”? BOINC Manager set to “run according to preferences”?

All checks out OK, but I think I may have a clue. A month or so ago ago I went to up the cache and the system said there was no user of that name and wouldn't let me in. I re-registered with Boinc using the same name but it alloccted me a new account number, and I cant get at the old account at all, so I cant merge them. You can see both accounts on the team I belong to.

Is it possible that Boinc is thinking that its running twice on the same computer and allocating all the resources to the "old" account? Perhaps the only way forward is to detach from the project, remove Boinc entirely and re-install it, then register with a different name. Before all this happened it was quite happily downloading a weeks worth of units.

---

Ah, Muriel Miggins - will she ever be mine

running 99.979%
working allowed 4.6481%
CPU 0.99979
duration factor .343873

OK the problem appears to be with the 4% work allowed time that appears. How has that happened and how can I fix it?. Other computers I have on Boinc are up at the 90% mark.

You could try to manually adjust that value in the client_state.xml file, located in \\program files\\boinc.

If you want to take the risk of messing with the boinc system files and maybe messing something up, here's the steps to take:

- shut down BOINC. (alltogether, check your task manager to be sure)
- backup your BOINC folder
- edit the client_state.xml file, located in your BOINC folder
- look for the line <active_frac>0.046481</active_frac> (your value could be different)
- change this value in something higher like 0.746481 (make sure to keep below 1, and use same number of digits)
- save the changes
- restart BOINC

Greetings,
Sander

---

Well it seems to have worked. I've just logged into my BOINC account and the active_frac has altered to 0.746481 which I set it to. I'll leave it for a day or so to settle down and see what work downloads I get, then maybe try changing it to 0.946481.

Many thanks to both of you for your help on this one, it was driving me nuts!

Happy Crunching

- shut down BOINC. (alltogether, check your task manager to be sure)
- backup your BOINC folder
- edit the client_state.xml file, located in your BOINC folder
- look for the line <active_frac>0.046481</active_frac> (your value could be different)
- change this value in something higher like 0.746481 (make sure to keep below 1, and use same number of digits)
- save the changes
- restart BOINC
Greetings,
Sander

---

Ah, Muriel Miggins - will she ever be mine

Is your “work while computer is in use” preference set to “yes”? BOINC Manager set to “run according to preferences”?

All checks out OK, but I think I may have a clue. A month or so ago ago I went to up the cache and the system said there was no user of that name and wouldn't let me in. I re-registered with Boinc using the same name but it alloccted me a new account number, and I cant get at the old account at all, so I cant merge them. You can see both accounts on the team I belong to.

Is it possible that Boinc is thinking that its running twice on the same computer and allocating all the resources to the "old" account? Perhaps the only way forward is to detach from the project, remove Boinc entirely and re-install it, then register with a different name. Before all this happened it was quite happily downloading a weeks worth of units.

Your name is irrelevant to Boinc accounts your email address is the unique identifier.

---

Boinc V7.2.42
Win7 i5 3.33G 4GB, GTX470

Despite having loads of disk space and the cache set to 10 days, I can only ever download 1 days work as Boinc keeps telling me the computer is over committed. I'm crunching a unit in about 3h 45m so 10 days cache should give me about 60 units, whereas I only get about 6 or 7 at a time.

Any ideas anyone?

One thing I noticed when I was playing with the "connect every 'x' days" setting:

(and this will change now that we have enhanced)

Standard work units have a 14 day deadline. The BOINC scheduler gets nervous when twice the connect interval is over the deadline.

Setting "connect every 'x' days" to 6 makes twice the deadline 12 days, and BOINC can likely meet a 14 day deadline by the second connection.

If it is worried about deadlines, adding work to a late queue does not help, so it suspends work-fetch.

Enhanced changes this because we have some work with a 4 1/2 day deadline, and some with a 50 day deadline, depending on the angle range.

---

Ned is right and I have stated also that if your connect time is set high that it can prevent you from getting work from seti.

If the work on the seti d/l server at the time of your request has a short deadline then you will not be issued work because the scheduler thinks that you will miss the deadline with your 10 day connection interval.

Short work = 4.34 days to deadline.

Therefore I would suggest a cache setting less than 4 days to decrease your chances of being refused work.
:)

---

98SE XP2500+ @ 2.1 GHz Boinc v5.8.8

And God said"Let there be light."But then the program crashed because he was trying to access the 'light' property of a NULL universe pointer.

Despite having loads of disk space and the cache set to 10 days, I can only ever download 1 days work as Boinc keeps telling me the computer is over committed. I'm crunching a unit in about 3h 45m so 10 days cache should give me about 60 units, whereas I only get about 6 or 7 at a time.

Any ideas anyone?

One thing I noticed when I was playing with the "connect every 'x' days" setting:

(and this will change now that we have enhanced)

Standard work units have a 14 day deadline. The BOINC scheduler gets nervous when twice the connect interval is over the deadline.

Setting "connect every 'x' days" to 6 makes twice the deadline 12 days, and BOINC can likely meet a 14 day deadline by the second connection.

If it is worried about deadlines, adding work to a late queue does not help, so it suspends work-fetch.

Enhanced changes this because we have some work with a 4 1/2 day deadline, and some with a 50 day deadline, depending on the angle range.

This has changed with 5.4.9.

1) The check for entering EDF if the deadline is within 2* connect every X is gone.
2) the check for enterint EDF if the deadline is within 24 hours is gone.
3) Instead, there is a test for work complete by round robin within report deadline - (connect every X + switch every Y + 24 hours). The result of this calculation is the computation deadline.

However, a machine in EDF is still not going to fetch more work.

---

BOINC WIKI