it used to take 3 hours to crunch a wu

Message boards : Number crunching : it used to take 3 hours to crunch a wu
Message board moderation

To post messages, you must log in.

Previous · 1 · 2

AuthorMessage
Pepperammi

Send message
Joined: 3 Apr 99
Posts: 200
Credit: 737,775
RAC: 0
United Kingdom
Message 41480 - Posted: 30 Oct 2004, 16:19:30 UTC - in response to Message 41438.  


> C Betteridge, thank you as well. Your suggestion is the one that Admiral
> proposed

Almost the same but the one i meant is setting screensaver to 'blank' instead of 'none'. justs give you the blank black screen.
thought you may prefer this as it clears the screen like boinc is suppose to when set to instead of leaving the boinc graph or the desktop on all the time.

al long as everything is sorted it doesn't realy matter.
ID: 41480 · Report as offensive
Profile xilef

Send message
Joined: 18 Jul 01
Posts: 43
Credit: 17,485,094
RAC: 0
United States
Message 41536 - Posted: 30 Oct 2004, 20:23:26 UTC - in response to Message 41480.  

Ah, yes. I see the difference now. Makes good sense. Thanks again.

> Almost the same but the one i meant is setting screensaver to 'blank' instead
> of 'none'. justs give you the blank black screen.
> thought you may prefer this as it clears the screen like boinc is suppose to
> when set to instead of leaving the boinc graph or the desktop on all the
> time.
>
> al long as everything is sorted it doesn't realy matter.
>
ID: 41536 · Report as offensive
Profile MarkRH
Avatar

Send message
Joined: 7 Dec 03
Posts: 32
Credit: 211,105
RAC: 0
United States
Message 41544 - Posted: 30 Oct 2004, 21:00:01 UTC - in response to Message 41474.  

> Normaly i only run boinc when im not on the pc cus i don't like the slowdown
> that it does have somtimes so i was amazed that STILL it took exactly 5hours
> and 4minutes;32secs... with all the other stuff it should have took much
> longer.
>
> i'm wondering if part the problem is seti have accidently some sort of
> processing limit. unlickly cus it always seems take all the CPU but how is all
> this possible?
>

Well, it's not all that surprising... most everyday tasks on the computer hardly use much of the CPU. Playing MP3 files on my system doesn't even register a percent of CPU usage (Winamp shows 0 CPU usage currently). The soundcard is doing all the work in this case. Most other things only cause momentary spikes in CPU usage. That's why if it wasn't for something like BOINC running, the System Idle Process is going over 95% most of the time.

Mark H.


ID: 41544 · Report as offensive
Pepperammi

Send message
Joined: 3 Apr 99
Posts: 200
Credit: 737,775
RAC: 0
United Kingdom
Message 41562 - Posted: 30 Oct 2004, 22:00:17 UTC - in response to Message 41544.  


> Well, it's not all that surprising... most everyday tasks on the computer
> hardly use much of the CPU. Playing MP3 files on my system doesn't even
> register a percent of CPU usage (Winamp shows 0 CPU usage currently). The
> soundcard is doing all the work in this case. Most other things only cause
> momentary spikes in CPU usage. That's why if it wasn't for something like
> BOINC running, the System Idle Process is going over 95% most of the time.

I completely acknowledge that. I know a lot of the things i was doing wouldn't have made any difference, but there were things that a was doing at times that should have. like i said i did some (about 45 min) photo editing in photoshop because my sis wonted some diffrent effects on her wedding pics. I'm no expert but i know that some of the processes i used are memory and CPU hungry, specialy on large files.
Most of all for me was, i use dual monitor and watch music vids more than the mp3's (i use visualisation with those too) and i know video files can take a steady 15-30%+ depending on the quality.

I wasn't expectin much but i still think it should have added at least 30-60min considering some of those apps. i might try waiting again and set up a simple (but long and hungry) 3D render. that realy would make it interesting.


ID: 41562 · Report as offensive
Ron Roe
Volunteer tester

Send message
Joined: 28 Feb 02
Posts: 156
Credit: 24,124
RAC: 0
United States
Message 41569 - Posted: 30 Oct 2004, 22:43:48 UTC - in response to Message 41562.  

The CPU Time of a WU is not the elapsed system time. It is the time the boinc application gets to execute on the CPU and is independent of CPU load.

Regards : )
ID: 41569 · Report as offensive
Profile MarkRH
Avatar

Send message
Joined: 7 Dec 03
Posts: 32
Credit: 211,105
RAC: 0
United States
Message 41571 - Posted: 30 Oct 2004, 22:56:44 UTC - in response to Message 41569.  

> The CPU Time of a WU is not the elapsed system time. It is the time the boinc
> application gets to execute on the CPU and is independent of CPU load.
>
> Regards : )
>

That is correct. I was going on the assumption that he was talking about the amount of real time/system time that had passed to complete the work, but it might have been an incorrect one. But, yes, if it takes an average of 5 hours to complete a WU and some process of higher priority steals the CPU for 30 minutes during the middle of processing a WU, it will take the system 5 hours and 30 minutes of real time to finish the WU. But, the CPU will still have only spent the same average of 5 hours actually working on the WU.

Mark H.
ID: 41571 · Report as offensive
Profile Steve Withers
Avatar

Send message
Joined: 25 Jun 99
Posts: 52
Credit: 3,083,069
RAC: 0
New Zealand
Message 41583 - Posted: 30 Oct 2004, 23:56:48 UTC - in response to Message 41058.  


> ... Only for Windows and possibly only for some of Windows.
>
> Meanwhile, Linux is working fine and fast!

Agreed.

My 5 Linux systems just crunch away at optimal speeds - all day, every day.


<img src=\"http://boinc.mundayweb.com/seti2/stats.php?userID=1248&trans=off\">
ID: 41583 · Report as offensive
Previous · 1 · 2

Message boards : Number crunching : it used to take 3 hours to crunch a wu


 
©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.