Screen Saver VS Full Tilt

Message boards : Number crunching : Screen Saver VS Full Tilt
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile 52 Aces
Avatar

Send message
Joined: 7 Jan 02
Posts: 497
Credit: 14,261,068
RAC: 67
United States
Message 931679 - Posted: 7 Sep 2009, 15:27:30 UTC

I'm running Win 7, and having trouble comparing WUpples to WUpples as I imagine inside each WU there is a mix of busy and quiet moments, but for anyone who has instrumented and optimized S@H and knows for sure:

Q: Is it faster to disable screensaver mode in the OS and just let the Boinc app crank in foreground?


The reason I ask, is I'm seeing the Screensaver with all of it's graphics slightly edging out the progress over time versus foreground mode, and I'm a bit surprised (although I can see the OS putting lots of other stuff to bed during screensaver).

Final detail is the WU I'm looking at is an 8 meg AP (my first, ETA was 595 hours, I'm going for it, but trying to trim a a few days in case I get into a pinch later (I'll be flying snd machine will have a fair amount of off time)).
ID: 931679 · Report as offensive
Alinator
Volunteer tester

Send message
Joined: 19 Apr 05
Posts: 4178
Credit: 4,647,982
RAC: 0
United States
Message 931682 - Posted: 7 Sep 2009, 15:46:48 UTC
Last modified: 7 Sep 2009, 15:47:49 UTC

It's absolutely faster to kill the screensaver.

In fact, the fastest of the optimized apps don't even include the graphics module in them.

A couple reasons for the graphics app running faster in screensaver mode is there is less competition for CPU time (obviously), as well as less competition for L2 and L3 cache space. Having to flush and reload cache space and wait around for main memory can be a real time killer. ;-)

Alinator
ID: 931682 · Report as offensive
John McCallum
Avatar

Send message
Joined: 5 Dec 04
Posts: 877
Credit: 599,458
RAC: 8
United Kingdom
Message 931685 - Posted: 7 Sep 2009, 15:52:52 UTC
Last modified: 7 Sep 2009, 15:56:01 UTC

I would not pay any attention to that time to compleation if I were you
my fist AP WU said something like 790 hours and a little later something like 1200 hours finished in 200 plus hours its just BOINC being just a bit conservative as regards time.Basicaly ignore it and let it do it's thing.OH and the best screen saver in the world is the off button on the monitor if running a desktop.
Old enough to know better(but)still young enough not to care
ID: 931685 · Report as offensive
Alinator
Volunteer tester

Send message
Joined: 19 Apr 05
Posts: 4178
Credit: 4,647,982
RAC: 0
United States
Message 931686 - Posted: 7 Sep 2009, 16:09:18 UTC
Last modified: 7 Sep 2009, 16:12:24 UTC

Agreed, the TDCF for AP is not the same as for MB.

You can get a better estimate of how long the task will take after you get some run time on it by multiplying the current CPU time used (in seconds) by the inverse of the current progress (expressed as a fraction). This makes BV a better choice to use for this since it will convert to seconds automatically for you. An HP scientific calculator would be my second choice! :-)

If you subtract the current CPU time from that, you get the time to go.

This has always worked well for me when the computed values in BM or BV were based on a inaccurate or just plain wrong TDCF. ;-)

Alinator
ID: 931686 · Report as offensive
1mp0£173
Volunteer tester

Send message
Joined: 3 Apr 99
Posts: 8423
Credit: 356,897
RAC: 0
United States
Message 931687 - Posted: 7 Sep 2009, 16:11:28 UTC - in response to Message 931679.  

I'm running Win 7, and having trouble comparing WUpples to WUpples as I imagine inside each WU there is a mix of busy and quiet moments, but for anyone who has instrumented and optimized S@H and knows for sure:

Q: Is it faster to disable screensaver mode in the OS and just let the Boinc app crank in foreground?

I'm pretty sure the screen saver is a separate program, and if running will compete with the science application for clock cycles.

I know that one of the optimizations is to remove the code that passes information to the screen saver, and as I remember, it was worth a couple of percent.

The percent completion is an estimate, as is the completion time. It will vary.

Alot.

ID: 931687 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 931766 - Posted: 7 Sep 2009, 19:19:48 UTC

In the days of classic, my first WU ever was on a Celeron 633, and with the screen saver turned on, the WU finished in a cool 119 hours. The next WU, I turned the screen saver off, and it finished in a more respectable 14 hours.

Ever since then, I have never enabled the screen saver, both in Classic and BOINC, and now with opti apps, you can't even use the SS.
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 931766 · Report as offensive
Josef W. Segur
Volunteer developer
Volunteer tester

Send message
Joined: 30 Oct 99
Posts: 4504
Credit: 1,414,761
RAC: 0
United States
Message 931889 - Posted: 8 Sep 2009, 6:27:53 UTC - in response to Message 931766.  

In the days of classic, my first WU ever was on a Celeron 633, and with the screen saver turned on, the WU finished in a cool 119 hours. The next WU, I turned the screen saver off, and it finished in a more respectable 14 hours.

Ever since then, I have never enabled the screen saver, both in Classic and BOINC, and now with opti apps, you can't even use the SS.

These days, most video cards provide OpenGL functions which make the application graphics use much less CPU. So even those with video cards insufficient to crunch on the GPU get some possible productivity increase from the newer technology. The fastest crunching is non-graphic but looking at the graphics occasionally would have little impact. I've never used a screensaver, in fact the original billing of Classic S@H as a screensaver kept me from participating for about 5 months, then I ran across something mentioning it could be run in the background without graphics and I was soon hooked.
                                                                Joe
ID: 931889 · Report as offensive
Profile Searcher
Avatar

Send message
Joined: 26 Jun 99
Posts: 139
Credit: 9,063,168
RAC: 15
United States
Message 932209 - Posted: 10 Sep 2009, 3:24:09 UTC

choose "blank" as the screen saver.
and set the power savings also.
:)
ID: 932209 · Report as offensive

Message boards : Number crunching : Screen Saver VS Full Tilt


 
©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.