Deadlines

copycat
copycat
Joined: 5 May 05
Posts: 18
Credit: 2113059
RAC: 0
Topic 189479

The deadlines E@H is expecting from me are ridiculously short, when comparing them to S@H's. Also, while S@H can manage 34,27 %/hour, E@H is only capable of 8,77 %/hour. The way I see it, you only manage the deadlines either if you only have E@H, or if you're having your pc BOINC-ing 24/24 7/7. I don't do that. During normal weekdays it's on 2-3 hours, on Friday 4-7 hours, not on Saturday and Sunday. In a (vain) attempt to be able to reach the current deadline I had it BOINC yesterday, and today, however, with only doing 1 hour worth of work in 2 hours time, it's useless. Also, in my results I see there's credits 'pending', and even some 'granted', However, when I go to the pending-menu, there's nothing there. Cuuld that be because I failed to meet the deadline on that previous WU? I've just alloted 66,67% of resource-time to E@H, while S@H only gets 33,33 % (S@H is already delivering completed WU's far ahead of the deadline anyway), and now I've also fiddled with the application switching-time. I am not planning on giving E@H this 'special treatment' every weekend you know.


Blank Reg
Blank Reg
Joined: 18 Jan 05
Posts: 228
Credit: 40599
RAC: 0

Deadlines

I think you need to stick to Seti@h or PP@H, for they have short crunch times.......

MarkF
MarkF
Joined: 12 Apr 05
Posts: 393
Credit: 1516715
RAC: 0

Based on your cpu type you

Based on your cpu type you should be able to crunch an e@h wu in something like 21,000 cpu-seconds. That works out to about 50 cpu-minutes day to finish within the deadline. Even with just 15 hours of up time a week you should be able to finish in time.
Sounds to me like something else is using up your available cpu cycles.

gravywavy
gravywavy
Joined: 22 Jan 05
Posts: 392
Credit: 68962
RAC: 0

RE: I am not planning on

Quote:
I am not planning on giving E@H this 'special treatment' every weekend you know.

I agree totally: nobody should give E@h special treatment, nor any other BOINC project. The whole point is to use spare cpu cycles, not run the machine especially.

(Exception: BOINC-hobbyists who get satisfaction from the very fact of delivering high throughput, but they'll be on 25/24 and 8/7 as soon as they figure out how...).

Two points:

First, the question of deadlines has been done to death in various threads. The project has considered these points before and are not likely to change because of one more posting, though change may happen after a new server is installed

Second, if I am right you are running in Linux. Sadly, E@h seems to perform particuallry badly in Linux. You will up to double your performance if you are willing to install Wine and run the Windows version of E@h under Wine/Linux. Sad but true, and much more so for E@h than on other projects. Lots more about this in this thread from "Cruncher's Corner". If you drop E@h and go to another project, then I wouldn't bother with Wine.

And finally, if you do decide to move on to another project, total respect to you for the work you have done for E@h while here. Every little helps. Nobody will be offended if you feel E@h is not for you, all things considered

If you don't have strong preferences for a particular scientific objective, then I'd recommend Predictor for any machine that is on <35 hr/wk - nice short wu that even my 700MHz PII can do in a single evening - and ask you also to try Orbit@home after it finishes beta (a long time away yet as it is not even in alpha testing yet).

~~gravywavy

Bronco
Bronco
Joined: 22 Jun 05
Posts: 6
Credit: 73429
RAC: 0

It's really desapointing to

It's really desapointing to see that BOINC projects are not able to work together. Actually, the problem of E@H deadlines is that you can only work on this project giving him 66% of your CPU, or even more.

I don't expect a miracle, but if running E@H, S@H and CPDN together is not possible, I'm going to leave E@H because it's the one guilty for the mess. And giving results after the dead line gives no sense. It means that the job done is not usefull, or that the deadlines are stupid.

Please, do something about that. You need big WU to calm down the servers, not short deadlines. And if you don't get back the results before deadline, you are obliged to resend the WU, which mean more work for nuts !

Keck_Komputers
Keck_Komputers
Joined: 18 Jan 05
Posts: 376
Credit: 5744955
RAC: 0

Bronco: Update to the 4.45

Bronco:
Update to the 4.45 version. It handles multiple projects much better than earlier versions. No need to set E@H to such a high percent it will just avoid downloading more work from the project from time to time.

BOINC WIKI

BOINCing since 2002/12/8

Bronco
Bronco
Joined: 22 Jun 05
Posts: 6
Credit: 73429
RAC: 0

RE: Bronco: Update to the

Message 13757 in response to message 13756

Quote:
Bronco:
Update to the 4.45 version. It handles multiple projects much better than earlier versions. No need to set E@H to such a high percent it will just avoid downloading more work from the project from time to time.


I am in 4.45, and it's not working fine. It may be due to Threadmaster, which effects may not be included in the computation to determine the amount of WU sent (Particularly on my laptop where the amount of CPU given is only 30%, for an Athlon 64 3000+)

The fact is that seti gives twice the time for WU that are half of E@H ones. I would say that working the same way with seti's deadlines should solve my problem, and other's too.

John McLeod VII
John McLeod VII
Moderator
Joined: 10 Nov 04
Posts: 547
Credit: 632255
RAC: 0

RE: RE: Bronco: Update to

Message 13758 in response to message 13757

Quote:
Quote:
Bronco:
Update to the 4.45 version. It handles multiple projects much better than earlier versions. No need to set E@H to such a high percent it will just avoid downloading more work from the project from time to time.

I am in 4.45, and it's not working fine. It may be due to Threadmaster, which effects may not be included in the computation to determine the amount of WU sent (Particularly on my laptop where the amount of CPU given is only 30%, for an Athlon 64 3000+)

The fact is that seti gives twice the time for WU that are half of E@H ones. I would say that working the same way with seti's deadlines should solve my problem, and other's too.


Because E@H deadlines are so short, 4.45 will work on those until they are done, and then not down load any E@H for a time to let the other projects catch up. Just let it run for a couple of weeks and watch to see what it does.

Bronco
Bronco
Joined: 22 Jun 05
Posts: 6
Credit: 73429
RAC: 0

RE: RE: RE: Bronco: Upd

Message 13759 in response to message 13758

Quote:
Quote:
Quote:
Bronco:
Update to the 4.45 version. It handles multiple projects much better than earlier versions. No need to set E@H to such a high percent it will just avoid downloading more work from the project from time to time.

I am in 4.45, and it's not working fine. It may be due to Threadmaster, which effects may not be included in the computation to determine the amount of WU sent (Particularly on my laptop where the amount of CPU given is only 30%, for an Athlon 64 3000+)

The fact is that seti gives twice the time for WU that are half of E@H ones. I would say that working the same way with seti's deadlines should solve my problem, and other's too.


Because E@H deadlines are so short, 4.45 will work on those until they are done, and then not down load any E@H for a time to let the other projects catch up. Just let it run for a couple of weeks and watch to see what it does.


OK, I'll try it after my holydays
Thanks

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.