Message boards :
Number crunching :
KWSN Windows optimized science apps - Share your results and problems!
Message board moderation
Previous · 1 . . . 6 · 7 · 8 · 9 · 10 · 11 · 12 . . . 13 · Next
Author | Message |
---|---|
KWSN - Chicken of Angnor Send message Joined: 9 Jul 99 Posts: 1199 Credit: 6,615,780 RAC: 0 |
I'm in contact with someone who told me that there are a number of functions that could be replaced by quicker counterparts from IPP's collection - hopefully, I'll get to look at his source code soon. Thanks for the Hint Alex :o) Oh, and my MacMega, er MacMini now pulls around 280 RAC...and its DCF is way lower than on my PCs, so again hats off to you. 1.4 GHz G4 laying waste to a 3 GHz P4 is pretty nice work indeed (both running optimized clients). Dr., I'm almost done running your latest failed WU (02ap05aa.8758.30433.711094.3.112) on my machine. So far, no errors - I'm using the exact same app you were (and running it through a debugger to catch any errors if they do come up). I'll let you know how it went. Regards, Simon. Donate to SETI@Home via PayPal! Optimized SETI@Home apps + Information |
nick Send message Joined: 22 Jul 05 Posts: 284 Credit: 3,902,174 RAC: 0 |
ok i have via C3 CPU that has MMX and SSE on it does it matter if i use the intel only one or should i use the AMD one app? nick |
KWSN - Chicken of Angnor Send message Joined: 9 Jul 99 Posts: 1199 Credit: 6,615,780 RAC: 0 |
You should use the generic SSE app - it runs on any PC, and has run successfully on a Via C3. Use this download link to grab it if you're running Windows, and this one if you're running Linux. HTH, Simon. Donate to SETI@Home via PayPal! Optimized SETI@Home apps + Information |
nick Send message Joined: 22 Jul 05 Posts: 284 Credit: 3,902,174 RAC: 0 |
You should use the generic SSE app - it runs on any PC, and has run successfully on a Via C3. ok thanks nick |
Clyde C. Phillips, III Send message Joined: 2 Aug 00 Posts: 1851 Credit: 5,955,047 RAC: 0 |
I'm really pleased with Simon's cruncher. I see the RAC of the main machine gradually climb toward 750; it had been below 600 all the time with the default. I did see one rotten apple with one unit that took almost four hours to crunch versus 2-1/2 for other units like it. I can't explain that anomaly because I haven't been playing Internet music lately. The 13-credit units (AR above 1.12 degree) yield best with my machine, the 60-credit units (around 0.4 degree) come next, and the 30-35 credit ones lag - these are the 0.8-or-so AR units. I have yet to see a VLAR unit crunched with Simon's cruncher. Meanwhile the subordinate machine lies apart waiting for that replacement nVidia GEForce 6200, due here July 20. |
Misfit Send message Joined: 21 Jun 01 Posts: 21804 Credit: 2,815,091 RAC: 0 |
Most early Prescotts are quicker with SSE2 than SSE3. That yours is also quicker using generic SSE2 than P4-specific SSE2 is interesting, usually it's the other way around. Since I already know this is a socket 478 I guess I had better switch to sse2. me@rescam.org |
nick Send message Joined: 22 Jul 05 Posts: 284 Credit: 3,902,174 RAC: 0 |
You should use the generic SSE app - it runs on any PC, and has run successfully on a Via C3. ok will give the times for the C3 when it finishes. nick |
KWSN - Chicken of Angnor Send message Joined: 9 Jul 99 Posts: 1199 Credit: 6,615,780 RAC: 0 |
Dr., the WU that failed on your computer crunched successfully on my Athlon64 3500+ without error - so at this point, I think it's not a problem directly related to the optimized app itself, but rather a corollary system problem like thermals or memory when the computer is under full load. When you use an optimized app, it stresses your system more than the default one would (even though they both show 100% CPU usage). It's like with cars - when you use Nitrous, they go faster, but they also break more easily, even though you can use the same engine for both. Have you ever done a "burn-in" with Prime95 or the like? What that means is use an application that runs lots of math (like the S@H apps do also) or other CPU-intensive tasks continuously and checks for errors doing so during a prolonged period of time. Say, 12 hours minimum, better would be 24-48. If it can get through that without any errors, I'd be surprised - of course, I may be wrong here, but again: on the same hardware using the same software, no crash for me. Regards, Simon. Donate to SETI@Home via PayPal! Optimized SETI@Home apps + Information |
KWSN - Chicken of Angnor Send message Joined: 9 Jul 99 Posts: 1199 Credit: 6,615,780 RAC: 0 |
Dr., or any other person who had trouble with the apps, would you be willing to try and crunch the same WU offline again? Would take ~3-4 hours, probably, it took 3 hrs 10 minutes on my computer with the debugger in the background. You could also try and use a temperature monitor to see whether your CPU has thermal trouble when you're crunching during that run. I've got a script that tests just that WU with the SSE2 client, double-clicking is all you need to do. Let me know :o) Regards, Simon. Donate to SETI@Home via PayPal! Optimized SETI@Home apps + Information |
littlegreenmanfrommars Send message Joined: 28 Jan 06 Posts: 1410 Credit: 934,158 RAC: 0 |
On my 4th day of crunching with Simon's optimised app. Speed has increased to almost double on some WUs! Biggest problem, if I were the competitive sort is that I'm returning WUs so much faster than other members of the quorum that I'm now building up a sizeable queue of results awaiting validation. Score is rising rapidly, RAC is dropping, after an initial huge rise, due to the number of results still awaiting validation. This will change, as other quorum members return their results. Again, awesome work, Simon! |
KWSN - Chicken of Angnor Send message Joined: 9 Jul 99 Posts: 1199 Credit: 6,615,780 RAC: 0 |
Thanks littlegreenmanfrommars! Win some, you lose some ;) I've also noticed my pending credit increase again and the DCF getting lower on all machines I'm running it on. As to the WU that bombed out on Dr.'s PC using the generic SSE2 app, it's now been crunched by two different AMD systems without problems, so I really have to assume the problem is with the machine it runs on. Dr., if you want, I can tell you how to do a "burn-in" that should make sure your puter works 100% fine, or you could, if you want to, crunch the WU again offline to see whether you have the same error again. Regards, Simon. Donate to SETI@Home via PayPal! Optimized SETI@Home apps + Information |
suf Send message Joined: 22 May 99 Posts: 39 Credit: 960,732 RAC: 0 |
Great thanks to you, Simon! I have two computers, and I have now swapped to your Win SSE3 and Win SSE2 respectively - both machines seem to be happy enough. |
LucaB76 - BOINC.Italy Send message Joined: 15 Feb 00 Posts: 6 Credit: 73,136 RAC: 0 |
Lol, that's funny :o) Hi Simon, I'm still here to report that since the second wu crunched the real speed reported by your app was correct and everything is fine now! Please... don't ask me what's happened with that first result, ok? ;-) By switching form P4 SSE3 to generic SSE2 application on my first Prescott I've gained a 10% on credits/hour... cool! Thanks for your benchmarks tool! Luca B. |
littlegreenmanfrommars Send message Joined: 28 Jan 06 Posts: 1410 Credit: 934,158 RAC: 0 |
Thanks littlegreenmanfrommars! You're very welcome, Simon. BTW, it's easier to call me lgm! :) As an example of the improvement, I did a WU worth 62.08 creds on the standard app in 11 hours 34 mins With your optimised app, a WU worth 67.23 creds took 5 hours 54 mins |
KWSN - Chicken of Angnor Send message Joined: 9 Jul 99 Posts: 1199 Credit: 6,615,780 RAC: 0 |
Okay, this is for anyone having trouble with the optimized clients - computation errors in results done on BOINC where other clients didn't get any. Please install BoincLogX and make it log and keep your WUs and results. This way, it will be a lot easier to track down things - I won't have to search for the WU manually or spend a lot of time tracking down things. Then, when you have a failed WU, please email it to me at simon <squiggle> zadra.org together with some more detailed information from BoincLogX, please. Thanks! Simon. Donate to SETI@Home via PayPal! Optimized SETI@Home apps + Information |
AC Send message Joined: 22 Jan 05 Posts: 3413 Credit: 119,579 RAC: 0 |
Did I make a mistake of installing? I haven't been keeping up with my WUs but I think I completed at least three of them (w/5.15) but can't see them in my results. I'm using version 5.4.9, and I extracted the optimized 5.15 P-4 sse3 files to the boinc/projects/setiathome.berkeley.edu folder. Is this message showing the right info?: 7/17/2006 6:33:56 PM||Starting BOINC client version 5.4.9 for windows_intelx86 7/17/2006 6:33:56 PM||libcurl/7.15.3 OpenSSL/0.9.8a zlib/1.2.3 7/17/2006 6:33:56 PM||Data directory: C:\\Program Files\\BOINC 7/17/2006 6:33:56 PM||Processor: 2 GenuineIntel Mobile Intel(R) Pentium(R) 4 CPU 3.06GHz 7/17/2006 6:33:56 PM||Memory: 991.36 MB physical, 1.58 GB virtual 7/17/2006 6:33:56 PM||Disk: 74.52 GB total, [edited] GB free 7/17/2006 6:33:56 PM|SETI@home|URL: http://setiathome.berkeley.edu/; Computer ID: 2447440; location: home; project prefs: default 7/17/2006 6:33:56 PM||General prefs: from SETI@home (last modified 2006-06-27 18:59:25) 7/17/2006 6:33:56 PM||General prefs: no separate prefs for home; using your defaults 7/17/2006 6:33:56 PM||Local control only allowed 7/17/2006 6:33:56 PM||Listening on port 31416 7/17/2006 6:33:56 PM|SETI@home|Resuming task 04ap05aa.7473.13138.404820.3.25_2 using setiathome_enhanced version 515 7/17/2006 6:33:56 PM|SETI@home|Resuming task 04ap05aa.1136.26704.409648.3.151_0 using setiathome_enhanced version 515 7/17/2006 7:03:28 PM|SETI@home|Sending scheduler request to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi |
KWSN - Chicken of Angnor Send message Joined: 9 Jul 99 Posts: 1199 Credit: 6,615,780 RAC: 0 |
No - you probably unpacked the zip file directly into the projects/setiathome.berkeley.edu directory instead of first unpacking it, then moving all files from the directory it makes (but NOT the directory) there. The BOINC messages you pasted show that it does not know you have the optimized app installed - else, it would state something like this: 16.07.2006 22:29:56||Starting BOINC client version 5.4.9 for windows_intelx86 Note the line saying "Found app_info.xml; using anonymous platform" - that's what needs to be there. HTH, Simon. Donate to SETI@Home via PayPal! Optimized SETI@Home apps + Information |
AC Send message Joined: 22 Jan 05 Posts: 3413 Credit: 119,579 RAC: 0 |
No - Blast! Okay Simon, thanks for pointing that out for me. Yeah I extracted directly to setiathome.berkeley.edu instead of unpacking and then moving. :( Again, thanks. |
Steve Cressman Send message Joined: 6 Jun 02 Posts: 583 Credit: 65,644 RAC: 0 |
No - I just drag and drop the files I need from the zip file to the seti folder and never have a problem with subfolders being created. :) 98SE XP2500+ @ 2.1 GHz Boinc v5.8.8 And God said"Let there be light."But then the program crashed because he was trying to access the 'light' property of a NULL universe pointer. |
AC Send message Joined: 22 Jan 05 Posts: 3413 Credit: 119,579 RAC: 0 |
No - I just deleted the directory folder and manually reinstalled the files again, and got this message when I restarted: 7/18/2006 2:45:32 PM|SETI@home|Found app_info.xml; using anonymous platform So I guess all is well. :) |
©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.