Is there a way to stop Boinc from doing CPU benchmarking automatically?

Message boards : Number crunching : Is there a way to stop Boinc from doing CPU benchmarking automatically?
Message board moderation

To post messages, you must log in.

AuthorMessage
FrankH
Avatar

Send message
Joined: 25 Feb 04
Posts: 25
Credit: 18,183
RAC: 0
Sweden
Message 40269 - Posted: 26 Oct 2004, 10:15:21 UTC
Last modified: 26 Oct 2004, 10:17:56 UTC

Is there a way to stop Boinc from doing CPU benchmarking automatically?

I'm using the standard windows boinc client (not the service)...

The reason why I ask this is because the computing done on WU's doesn't affect my computer at all, if something runs, it just have the lower priority, and get less cpu time... so I let it always run, even when playing games etc. (with one exception, online gaming, as the issue I have will affect anyone close to me)

But, and a big but, when the CPU benchmark starts it's just like the priority have changed to normal or even high, and everything starts to get slowdowns, even my Hauppauge WinTV card get a choppy picture. And that minute of CPU benchmark have sofar not collided with online racing, but I feel it's only a matter of time before I forget to "suspend" boinc before loading up nr2003 or similar ...

so, is there a way to make manual cpu benchmark updates, or at least disable the CPU benchmarking on certain time of the day(like on evenings)?

(Edit: I have looked around but not found any documentation about this, either missed it or there isn't such a feature)
________<br>We\'re not living in a black and white world
ID: 40269 · Report as offensive
Ulrich Metzner
Volunteer tester
Avatar

Send message
Joined: 3 Jul 02
Posts: 1256
Credit: 13,565,513
RAC: 13
Germany
Message 40276 - Posted: 26 Oct 2004, 11:10:45 UTC

It does benchmarking about every five days. To be sure you're online session isn't troubled you can simply start the benchmarks manually and let them run thru before going online. Then you have 5 safe days ;)

Aloha, Uli

ID: 40276 · Report as offensive
FrankH
Avatar

Send message
Joined: 25 Feb 04
Posts: 25
Credit: 18,183
RAC: 0
Sweden
Message 40289 - Posted: 26 Oct 2004, 13:06:17 UTC - in response to Message 40276.  

Probably a good enough way to solve it

Thanks! :)
________<br>We\'re not living in a black and white world
ID: 40289 · Report as offensive
Profile Paul D. Buck
Volunteer tester

Send message
Joined: 19 Jul 00
Posts: 3898
Credit: 1,158,042
RAC: 0
United States
Message 40796 - Posted: 27 Oct 2004, 21:39:09 UTC

I think that when the benchmark tests are run, the tests are not run at the lower priority. That means that the benchmark process gets most of the CPU. You see this as a slow-down of all other programs.
ID: 40796 · Report as offensive
Walt Gribben
Volunteer tester

Send message
Joined: 16 May 99
Posts: 353
Credit: 304,016
RAC: 0
United States
Message 40802 - Posted: 27 Oct 2004, 22:03:07 UTC - in response to Message 40796.  

> I think that when the benchmark tests are run, the tests are not run at the
> lower priority. That means that the benchmark process gets most of the CPU.
> You see this as a slow-down of all other programs.
>

Paul,

BOINC doesn't raise the priority to run benchmarks, although that would probably make the numbers more consistent. Priority is left at what BOINC started at, "normal" (8), meaning it has to share the processor with everything else at that priority. It will certainly interfere with another CPU intensice task, just as other tasks running interfere with the benchmarks.

ID: 40802 · Report as offensive
Profile Paul D. Buck
Volunteer tester

Send message
Joined: 19 Jul 00
Posts: 3898
Credit: 1,158,042
RAC: 0
United States
Message 40821 - Posted: 27 Oct 2004, 23:12:16 UTC - in response to Message 40802.  

> Paul,
>
> BOINC doesn't raise the priority to run benchmarks, although that would
> probably make the numbers more consistent. Priority is left at what BOINC
> started at, "normal" (8), meaning it has to share the processor with
> everything else at that priority. It will certainly interfere with another
> CPU intensice task, just as other tasks running interfere with the
> benchmarks.

See what happens to my brain when I take a long trip across the country...

:)

Though I am surprised because it sure seemed to act that way on my systems ...

I will have to test this on a ccouple of systems to see if I can prove this one way or the 'tother (if nothing else, to satisfy my doubting thomas kind of mind). But, I see the logic of your statement, BOINC runs at normal, and the benchmark also runs ar normal, but the child science applications runs at the lower priorities ...
ID: 40821 · Report as offensive
Walt Gribben
Volunteer tester

Send message
Joined: 16 May 99
Posts: 353
Credit: 304,016
RAC: 0
United States
Message 40847 - Posted: 28 Oct 2004, 1:12:15 UTC - in response to Message 40821.  

> > Paul,
> >
> > BOINC doesn't raise the priority to run benchmarks, although that would
> > probably make the numbers more consistent. Priority is left at what
> BOINC
> > started at, "normal" (8), meaning it has to share the processor with
> > everything else at that priority. It will certainly interfere with
> another
> > CPU intensice task, just as other tasks running interfere with the
> > benchmarks.
>
> See what happens to my brain when I take a long trip across the country...
>
> :)
>
> Though I am surprised because it sure seemed to act that way on my systems
> ...
>
> I will have to test this on a ccouple of systems to see if I can prove this
> one way or the 'tother (if nothing else, to satisfy my doubting thomas kind of
> mind). But, I see the logic of your statement, BOINC runs at normal, and the
> benchmark also runs ar normal, but the child science applications runs at the
> lower priorities ...
>

I keep falling into the same trap, especially when I look at the difference in benchmark times between old BOINC and new BOINC. But I checked to make sure, especially as I get quite different results from benchmarks run standalone vs. running other programs.

Visual Studio comes with a process viewer, its pretty good in that it you can check the priority of each thread. That usually what I use to check priority.

If you don't have Visual Studio, you can get ProcessExplorer from SysInternals. Its free, get it here. Theres more good programs and utilities there, worth checking out.

You can use PE to raise or lower BOINC's priority "for testing" to see how they relate to the benchmark results and impact on other applications. Might be a good idea to suspend SETI first though, I had a few times where it ended early because I set BOINC too low. I use the estimates in the work tab for waiting WU's rather than the raw *stones to check differences.
ID: 40847 · Report as offensive
Profile Paul D. Buck
Volunteer tester

Send message
Joined: 19 Jul 00
Posts: 3898
Credit: 1,158,042
RAC: 0
United States
Message 40947 - Posted: 28 Oct 2004, 12:03:56 UTC - in response to Message 40847.  

> I keep falling into the same trap, especially when I look at the difference in
> benchmark times between old BOINC and new BOINC. But I checked to make sure,
> especially as I get quite different results from benchmarks run standalone vs.
> running other programs.
>
> Visual Studio comes with a process viewer, its pretty good in that it you can
> check the priority of each thread. That usually what I use to check
> priority.

You can use the process manager (CTL-ALT-DEL) to do the same thing in that you can see the default priority. All you have to do is to add the column to the display. I may try it here as soon as I can back up my daughter's computer.
ID: 40947 · Report as offensive
Pepo
Volunteer tester
Avatar

Send message
Joined: 5 Aug 99
Posts: 308
Credit: 418,019
RAC: 0
Slovakia
Message 40948 - Posted: 28 Oct 2004, 12:36:04 UTC - in response to Message 40947.  

> > Visual Studio comes with a process viewer, its pretty good in that it you
> > can check the priority of each thread. That usually what I use to check
> > priority.
>
> You can use the process manager (CTL-ALT-DEL) to do the same thing in that you
> can see the default priority. All you have to do is to add the column to the
> display. I may try it here as soon as I can back up my daughter's computer.
>
That's true, but sometimes the priority of particular thread(s) of the process can differ substantially (up or down) from the priority of other siblings or of the parent process. The Process Manager can't show this, Process Viewer and better ProcessExplorer (and many others) can.

It's probably not the case of Boinc manager (running with normal priority - 8), but e.g. the SETI client (running default with lowered priority - 4) contain threads with priorities, varying from 2 (science code?) to 6-8 (graphics?) to 15 (RPC?).

Peter
ID: 40948 · Report as offensive

Message boards : Number crunching : Is there a way to stop Boinc from doing CPU benchmarking automatically?


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.