Message boards :
Number crunching :
Optimised AP v5.00 - initial release
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · 6 . . . 9 · Next
Author | Message |
---|---|
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14655 Credit: 200,643,578 RAC: 874 |
Logan did a xml file for me so that I can do both AP and MB, so I have downloaded the file and see what is the difference, then see if I can save it as an xml file. Mine seems to open in outlook not notepad. Since you're using Windows, right-click on the file instead of double-clicking. From the fly-out menu, choose "Edit" (second item), if it's shown to you. If "Edit" isn't available (it doesn't show on all systems, for some reason), look for "Open with..." a bit further down, and try to find Notepad from there. |
[B^S] madmac Send message Joined: 9 Feb 04 Posts: 1175 Credit: 4,754,897 RAC: 0 |
Logan did a xml file for me so that I can do both AP and MB, so I have downloaded the file and see what is the difference, then see if I can save it as an xml file. Mine seems to open in outlook not notepad. OK so this new xml file will do both AP and MB optimised or will it do just one. I am running your version which does both Logan and I have open the new version and all I see is AP not MB |
[B^S] madmac Send message Joined: 9 Feb 04 Posts: 1175 Credit: 4,754,897 RAC: 0 |
|
Logan Send message Joined: 26 Jan 07 Posts: 743 Credit: 918,353 RAC: 0 |
I am sending my apologise to Logan I am half asleep been up since 3:00UTC I am a carer looking after my mother who has had a double bypass and a stroke. Even though I have sent him a private message thought I would do it on here as well. So I apologise again Logan for my mistake. What do you say? It's not needed, my friend... Only tell me if it works...:) Best regards from Barcelona. And give my best wishes to your mother. Logan. BOINC FAQ Service (Ahora, también disponible en Español/Now available in Spanish) |
[B^S] madmac Send message Joined: 9 Feb 04 Posts: 1175 Credit: 4,754,897 RAC: 0 |
|
Logan Send message Joined: 26 Jan 07 Posts: 743 Credit: 918,353 RAC: 0 |
Will do after my AP has finished and I restarted BOINC after placing it in my project folder. If you want, you can replace it now and forget... (stopping and restarting BOINC, of course (remember to add the new files from the dowload first)) I was do it before (to be sure) sent you the .xml and it works... .xml file is ready to continue your 4.36 wu. I'm crunching now one of that. Logan. BOINC FAQ Service (Ahora, también disponible en Español/Now available in Spanish) |
[B^S] madmac Send message Joined: 9 Feb 04 Posts: 1175 Credit: 4,754,897 RAC: 0 |
|
Logan Send message Joined: 26 Jan 07 Posts: 743 Credit: 918,353 RAC: 0 |
I have done it and I have not got any problems at this moment. I did exit and then restarted BOINC it said I'm glad to know it, my friend. That says 'all is fine...' The new indicator 500 in AP WU's only will assigned when you dowload new AP wu's. Let me know the progress, please. Logan. BOINC FAQ Service (Ahora, también disponible en Español/Now available in Spanish) |
Cosmic_Ocean Send message Joined: 23 Dec 00 Posts: 3027 Credit: 13,516,867 RAC: 13 |
... I will have to concur with this observation. I've got three WUs that started with the new 4.37 opti: I was running 49-51 hours on the previous opti, and this one, with some extrapolation, is looking to run 55-62. Again, it's many times better than the old 135 hours I used to run, but there does appear to be a slight slowdown from the previous opti to the current one. Linux laptop: record uptime: 1511d 20h 19m (ended due to the power brick giving-up) |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14655 Credit: 200,643,578 RAC: 874 |
The first one in that list is task 1060056077, running on your 'Dual-Core AMD Opteron(tm) Processor 2210 [x86 Family 15 Model 65 Stepping 2]', right? It does seem as if that build has problems on AMD Opteron/Phenom CPUs (and possibly others, as yet unreported). By contrast, my test notes for an Intel Q6600 show a speedup from 18:32:47 down to 17:16:42 for the first run of that build. All goes to demonstrate that we need a larger and more cosmopolitan testing fleet than at present. |
Byron S Goodgame Send message Joined: 16 Jan 06 Posts: 1145 Credit: 3,936,993 RAC: 0 |
All goes to demonstrate that we need a larger and more cosmopolitan testing fleet than at present. Adding my two cents worth. My dual core E2180 which had a 4.36 going and was almost half way completed with the old opti app for AP, was looking around 42 hours (mid range of norm for it). Now after adding the new opti app I'm pretty sure it will closer to 44 now. Hard to say on that one since with the old opti I use to get those times every now and again. Am anxious to see what the two 5.00 that I have later on will be doing. My AMD64 3800+ was doing AP in around 39 hours on the old opti app, and with the new 5.00 app is looking at the moment like it will be closer to what my E2180 use to do at 42 and a half hours. Proof will be when they actually finish but the new opti app does appear to be slightly slower than the last opti app was at this point on these two boxes of mine. |
Ralph Send message Joined: 27 Mar 01 Posts: 34 Credit: 299,146 RAC: 0 |
My next project for the day was to download the new optimized app. The link in the message doesn't seem to work for me. I'll try again in the morning. |
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0 |
The apparent slowdown with 4.37 on some systems should be a very minor misfortune, as it will only be used to finish out a limited number of WUs which were marked to be processed by 4.x. If someone has a very large cache of such and a host on which 4.37 is slower it might make sense to replace the 4.37 with the prior 4.35. The 5.00 rev 69 build is certainly not the final Lunatics effort, it had sufficient testing that we were confident it was accurate, and it showed definite speed improvements on the test systems. As things settle out after the transition there will be single 5.xx application releases. With luck, the timing of those won't be forced by unannounced project changes so some of the many avenues for further improvement can be explored. Joe |
Byron S Goodgame Send message Joined: 16 Jan 06 Posts: 1145 Credit: 3,936,993 RAC: 0 |
My next project for the day was to download the new optimized app. Richard posted them again with a more direct link in this message |
Byron S Goodgame Send message Joined: 16 Jan 06 Posts: 1145 Credit: 3,936,993 RAC: 0 |
Just another update. The AP 4.36 that I put the new opti app on did complete with a cpu time of 44:16:38. That is at least a couple of hours more than I feel it would have completed by the previous opti but who can really say for sure. I'll leave the app info file as is and see if I get other 4.36 and their times. The AMD64 3800+ I may have spoken too early about, it is about to come up to 10 cpu hours and is over 27.082% completed so it's seems to be out of the 40 hour time now and more in line with the previous opti. |
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0 |
Just another update. The AP 4.36 that I put the new opti app on did complete with a cpu time of 44:16:38. That is at least a couple of hours more than I feel it would have completed by the previous opti but who can really say for sure. I'll leave the app info file as is and see if I get other 4.36 and their times. When you're using an app_info.xml, the Scheduler does not tell the core client to select a specific version of the application for new work. The core client merely selects the one with the highest version. So you won't get any new work for 4.36, it will all be assigned to 5.00. If you were not using an app_info.xml, the Scheduler would specify that any new work should be done with stock 5.00 and your host would download the app if it didn't already have it. Thus, new work is done with 5.00 either way. That may not be the best thing when the new work is a reissue intended to resolve an inconclusive pair done with 4.36, but there is no BOINC mechanism to handle that situation better. Had the project left work generation off for two or three weeks and not released 5.00, the number of unresolved WUs would have dropped to a low level. Then they could have released 5.00 and started splitting new work with much less confusion. They started out to do that, Matt noted the ap_splitters had been disabled to allow work to drain. The remainder of the process showed some Murphy's law effects. Joe |
MarkJ Send message Joined: 17 Feb 08 Posts: 1139 Credit: 80,854,192 RAC: 5 |
My next project for the day was to download the new optimized app. I used the link from the 1st message in the thread and that seemed to work for me yesterday. My machines have been set to no new work for Seti for a day or so. As they finished off their old AP work I have uploaded them, reported them, shutdown boinc, updated app_info, removed older apps, put new apps in place, restarted boinc, allowed new work and told it to update project. Repeat 5 times quickly to cover the entire farm. My first AP 5.00 work unit should be done in about 5 hours. BOINC blog |
MarkJ Send message Joined: 17 Feb 08 Posts: 1139 Credit: 80,854,192 RAC: 5 |
When you're using an app_info.xml, the Scheduler does not tell the core client to select a specific version of the application for new work. The core client merely selects the one with the highest version. So you won't get any new work for 4.36, it will all be assigned to 5.00. That begs the question then, why is there a 4.37 in the app_info and the achive? Is it just for those that may be in the middle of crunching an older work unit and want to update immediately? BOINC blog |
Byron S Goodgame Send message Joined: 16 Jan 06 Posts: 1145 Credit: 3,936,993 RAC: 0 |
When you're using an app_info.xml, the Scheduler does not tell the core client to select a specific version of the application for new work. The core client merely selects the one with the highest version. So you won't get any new work for 4.36, it will all be assigned to 5.00. If I were to complete the 5.00 tasks I have with the new opti app, as well as drain the rest of my cache to keep from having anything flushed if I make a mistake lol... If I then close down BOINC and reinstall the old opti app would I then be matched up only with the old 4.36 bunch of WU's out there, or mark ones that should be done by 5.00 as 4.36 where I would just be screwing things up more? Or would I even get AP WU's? |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
When you're using an app_info.xml, the Scheduler does not tell the core client to select a specific version of the application for new work. The core client merely selects the one with the highest version. So you won't get any new work for 4.36, it will all be assigned to 5.00. Correct, and a couple of extremely obscure bugs in 4.35 stock prevented knowing redistribution of the earlier 4.35 app in the package. Expect 4.37 to have a lifespan of no more than a few days, when a 'pure' v5 release (not aimed at the difficult transition period) will be changed, as the probability of someone having partially crunched 4.35/4.36 tasks becomes closer to zero. "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
©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.