Optimized Clients??

Message boards : Number crunching : Optimized Clients??
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · Next

AuthorMessage
Yamanipanuchi

Send message
Joined: 18 May 00
Posts: 87
Credit: 10,646,074
RAC: 0
United States
Message 435004 - Posted: 12 Oct 2006, 6:25:40 UTC

I've so far tried 2 diffrent optimized clients, on 4 diffrent PC's. I unfortunatly have seen no improovment in processing.

I can confirm both in boincview, and at command line that the the correct application is being used.

I've tried it on a PIII 700, PII 400(Dual), P4 1.8, P-D 3.4. I no noticable change in MFIO's or in overall time it takes to process the file.

I guess the question im asking is... Is this normal? Or should I see a noticable jump in measurable speed?
ID: 435004 · Report as offensive
Profile KWSN - Chicken of Angnor
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 9 Jul 99
Posts: 1199
Credit: 6,615,780
RAC: 0
Austria
Message 435016 - Posted: 12 Oct 2006, 7:50:35 UTC

Hi,

I checked your P-D 3.4 GHz, and it is indeed using an optimized app.

However, I also saw it is using calibration to adjust the CPU times, which is probably why you're not noticing much.

Please turn off calibration for S@H or use the default BOINC client to make your result stats accurately represent what you crunch :o)

HTH,
Simon.
Donate to SETI@Home via PayPal!

Optimized SETI@Home apps + Information
ID: 435016 · Report as offensive
Alinator
Volunteer tester

Send message
Joined: 19 Apr 05
Posts: 4178
Credit: 4,647,982
RAC: 0
United States
Message 435086 - Posted: 12 Oct 2006, 12:20:49 UTC
Last modified: 12 Oct 2006, 12:21:18 UTC

I took a look at the older ones and the last result reported for all of them seems to indicate they ran it with the stock app for some reason.

FWIW, I got about a 20-25 % improvement with both the MMX and SSE apps on PII/K6-x and PIII class machines.

Also, I thought you said you were going to turned calibrate off on your PD. ;-)

Alinator
ID: 435086 · Report as offensive
Odysseus
Volunteer tester
Avatar

Send message
Joined: 26 Jul 99
Posts: 1808
Credit: 6,701,347
RAC: 6
Canada
Message 435216 - Posted: 12 Oct 2006, 18:41:06 UTC - in response to Message 435004.  
Last modified: 12 Oct 2006, 18:43:52 UTC

I've so far tried 2 diffrent optimized clients, on 4 diffrent PC's. I unfortunatly have seen no improovment in processing.

Optimized clients make next to no difference in CPU times, which are almost entirely dependent on the science applications. On my Macs, even when BOINC Manager is open it uses only about 2% of the CPU, while the app(s) running under it get over 95% (subject to whatever else I may be doing at the time). That suggests that there’s very little to be gained by streamlining it. (BTW, I’ve noticed that BM uses around 2.5% when it’s on the Tasks pane, or the Projects pane while a ‘countdown’ is in progress, but less than 1.5% when on the other panes.)

Aside from extra control & monitoring features it may have, an optimized BOINC version is usually intended primarily to adjust credit claims, by altering either benchmark scores or reported processing times. This has little effect in projects like S@h and Einstein@home (actually none there—they determine credit server-side), whose crediting procedures are being weaned away from the default time-&-benchmark computation.
ID: 435216 · Report as offensive
Yamanipanuchi

Send message
Joined: 18 May 00
Posts: 87
Credit: 10,646,074
RAC: 0
United States
Message 435271 - Posted: 12 Oct 2006, 20:39:41 UTC

calibration for S@H

I've never seen this feature, or didnt know I set it. How would I go about shutting it off?

What numbers do any of you use to compare the Optimzed to normal? CPU times can always vary. So I not sure where the good measurement would be?
ID: 435271 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14653
Credit: 200,643,578
RAC: 874
United Kingdom
Message 435285 - Posted: 12 Oct 2006, 21:03:22 UTC - in response to Message 435271.  
Last modified: 12 Oct 2006, 21:04:01 UTC

calibration for S@H

I've never seen this feature, or didnt know I set it. How would I go about shutting it off?

What numbers do any of you use to compare the Optimzed to normal? CPU times can always vary. So I not sure where the good measurement would be?

You're running Trux's version of BOINC, same as me.

Look for the file 'truxoft_prefs.xml' in the BOINC directory (not any project sub-folder)

Open the file for editing (notepad will do). Look for the calibration line, and edit it to read:
    <calibrate_credit>off</calibrate_credit>

Save changes, close and restart BOINC.

People will be looking at the 'CPU time' column in your results to compare apps: Trux's calibration adjusts this figure - that's why it's unhelpful now it no longer affects credit claims.

ID: 435285 · Report as offensive
IHTIMAN SETI TEAM
Volunteer tester
Avatar

Send message
Joined: 8 Dec 01
Posts: 109
Credit: 1,532,174
RAC: 0
Bulgaria
Message 435299 - Posted: 12 Oct 2006, 21:27:39 UTC

From where i can download a optimized BOINC app ? :)
Dell XPS L702x
smokin' weed and searching for aliens .... ;]
TRUSTNO1
ID: 435299 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14653
Credit: 200,643,578
RAC: 874
United Kingdom
Message 435304 - Posted: 12 Oct 2006, 21:34:27 UTC - in response to Message 435299.  
Last modified: 12 Oct 2006, 21:35:25 UTC

From where i can download a optimized BOINC app ? :)

You can download optimised SETI apps from here.

You can't - or at least shouldn't - download optimised BOINC clients: mine is left over from a long time ago, and isn't really any use now (I'm just fond of the devil I know....)

[edit - typo: contradicted myself!]
ID: 435304 · Report as offensive
SETI User

Send message
Joined: 29 Jun 02
Posts: 369
Credit: 0
RAC: 0
Germany
Message 435308 - Posted: 12 Oct 2006, 21:38:22 UTC - in response to Message 435299.  
Last modified: 12 Oct 2006, 22:07:51 UTC

From where i can download a optimized BOINC app ? :)


@ Yamanipanuchi too

When you want to do more WUs in the same time -> optimized S@H applications!



Boinc- clients


S@H applications





Greetings!



ID: 435308 · Report as offensive
IHTIMAN SETI TEAM
Volunteer tester
Avatar

Send message
Joined: 8 Dec 01
Posts: 109
Credit: 1,532,174
RAC: 0
Bulgaria
Message 435315 - Posted: 12 Oct 2006, 21:49:49 UTC

Thanks , your are realy fast :))) I must use the Calibrated app or .... ?
Dell XPS L702x
smokin' weed and searching for aliens .... ;]
TRUSTNO1
ID: 435315 · Report as offensive
SETI User

Send message
Joined: 29 Jun 02
Posts: 369
Credit: 0
RAC: 0
Germany
Message 435319 - Posted: 12 Oct 2006, 21:56:35 UTC - in response to Message 435315.  
Last modified: 12 Oct 2006, 22:04:47 UTC

Thanks , your are realy fast :))) I must use the Calibrated app or .... ?



I use Boinc V5.3.12.tx36 !

But you know, that your speed of crunching will not change with an other Boinc- client...
...you will have only more service or functions...

You will have more speed only with optimized S@H applications!






ID: 435319 · Report as offensive
IHTIMAN SETI TEAM
Volunteer tester
Avatar

Send message
Joined: 8 Dec 01
Posts: 109
Credit: 1,532,174
RAC: 0
Bulgaria
Message 435329 - Posted: 12 Oct 2006, 22:23:58 UTC

I download it , put it in my boinc dir , and now the boinc manager dont start .... Some error with some librarys ....
Dell XPS L702x
smokin' weed and searching for aliens .... ;]
TRUSTNO1
ID: 435329 · Report as offensive
SETI User

Send message
Joined: 29 Jun 02
Posts: 369
Credit: 0
RAC: 0
Germany
Message 435343 - Posted: 12 Oct 2006, 22:44:11 UTC - in response to Message 435329.  
Last modified: 12 Oct 2006, 22:55:53 UTC

I download it , put it in my boinc dir , and now the boinc manager dont start .... Some error with some librarys ....



Before you copy the unpacked files in the Boinc folder you must copy your Boinc folder...

Close/ exit Boinc...
Then copy the unpacked files in the Boinc folder...
Restart Boinc...

I had done the "update" on my (original Berkeley) Boinc V5.2.13...

You cannot use Trux Boinc files (client) on a higher/ later client as V5.3.12 !
I think you can use latest only the original V5.2.13 for the Trux- update...



ID: 435343 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14653
Credit: 200,643,578
RAC: 874
United Kingdom
Message 435361 - Posted: 12 Oct 2006, 23:28:54 UTC - in response to Message 435343.  
Last modified: 12 Oct 2006, 23:32:56 UTC

You cannot use Trux Boinc files (client) on a higher/ later client as V5.3.12 !
I think you can use latest only the original V5.2.13 for the Trux- update...

Well, you can - I did it last week on a machine that's only crunching Einstein, none of the ones you can see here - but I don't see why you should - and you certainly shouldn't rush into doing it without pausing to understand what you're doing and why you're doing it.
ID: 435361 · Report as offensive
zombie67 [MM]
Volunteer tester
Avatar

Send message
Joined: 22 Apr 04
Posts: 758
Credit: 27,771,894
RAC: 0
United States
Message 435373 - Posted: 13 Oct 2006, 0:20:42 UTC

In windows, how do you tell what your processor supports (MMX/SSE/SSE2/SSE3)?
ID: 435373 · Report as offensive
Astro
Volunteer tester
Avatar

Send message
Joined: 16 Apr 02
Posts: 8026
Credit: 600,015
RAC: 0
Message 435375 - Posted: 13 Oct 2006, 0:23:52 UTC

CPUZ will tell you.
ID: 435375 · Report as offensive
Yamanipanuchi

Send message
Joined: 18 May 00
Posts: 87
Credit: 10,646,074
RAC: 0
United States
Message 435378 - Posted: 13 Oct 2006, 0:30:05 UTC

I honestly completely forgot I was using Truxsoft. I just removed it. But unfortunatly now Im geting a slower speed reported. TruxSoft had a command for CPU Affinity, I seems to get more more performance? Or was that just me?



zombie67

CPU-Z is a GREAT program for checking what type of processor you have.
ID: 435378 · Report as offensive
zombie67 [MM]
Volunteer tester
Avatar

Send message
Joined: 22 Apr 04
Posts: 758
Credit: 27,771,894
RAC: 0
United States
Message 435401 - Posted: 13 Oct 2006, 1:35:22 UTC

Thanks. I see that there is also a program at lunatics.at that will also tell you a lot about your processors.
ID: 435401 · Report as offensive
SETI User

Send message
Joined: 29 Jun 02
Posts: 369
Credit: 0
RAC: 0
Germany
Message 435544 - Posted: 13 Oct 2006, 9:27:04 UTC - in response to Message 435361.  
Last modified: 13 Oct 2006, 9:40:33 UTC

You cannot use Trux Boinc files (client) on a higher/ later client as V5.3.12 !
I think you can use latest only the original V5.2.13 for the Trux- update...

Well, you can - I did it last week on a machine that's only crunching Einstein, none of the ones you can see here - but I don't see why you should - and you certainly shouldn't rush into doing it without pausing to understand what you're doing and why you're doing it.





@ Richard Haselgrove

Thanks for your friendly contribution in this thread!

And which is the latest Boinc Version you can supplement with the Trux- files?

So that I can learn something here...




ID: 435544 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14653
Credit: 200,643,578
RAC: 874
United Kingdom
Message 435706 - Posted: 13 Oct 2006, 18:24:33 UTC - in response to Message 435544.  

And which is the latest Boinc Version you can supplement with the Trux- files?

So that I can learn something here...

This machine is the one I was describing - attached 2 October 2006: I used the currently recommended BOINC 5.4.11 to do the initial install.

I'm expecting a couple of Core 2s next week, which I may get a chance to 'burn in' for the client over next weekend - I'll see if the same trick works for alpha 5.6.5 (since they won't be permanent crunchers, alas), and report back.
ID: 435706 · Report as offensive
1 · 2 · 3 · Next

Message boards : Number crunching : Optimized Clients??


 
©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.