Message boards :
Number crunching :
Optimized Astropulse 5.00 rev103 for Windows
Message board moderation
Author | Message |
---|---|
Gecko Send message Joined: 17 Nov 99 Posts: 454 Credit: 6,946,910 RAC: 47 |
Lunatics has just released an upgrade to Astropulse 5.0 for Windows, rev103. This upgrade is optional, but does show a significant performance increase. Please see release post here: http://setiathome.berkeley.edu/forum_thread.php?id=31810&nowrap=true#856092 and @ Lunatics: http://lunatics.kwsn.net/index.php Thanks and Have fun! Lunatics Team |
Mike Davis Send message Joined: 17 May 99 Posts: 240 Credit: 5,402,361 RAC: 0 |
Lunatics has just released an upgrade to Astropulse 5.0 for Windows, rev103. Just a quick Q. <app_info> <app> <name>astropulse</name> </app> <file_info> <name>ap_5.00r103_SSE3.exe</name> <executable/> </file_info> <app_version> <app_name>astropulse</app_name> <version_num>500</version_num> <file_ref> <file_name>ap_5.00r103_SSE3.exe</file_name> <main_program/> </file_ref> </app_version> </app_info> Is the app_info with the file. <app_info> <app> <name>setiathome_enhanced</name> </app> <file_info> <name>AK_v8_win_x64_SSE41.exe</name> <executable/> </file_info> <app_version> <app_name>setiathome_enhanced</app_name> <version_num>603</version_num> <file_ref> <file_name>AK_v8_win_x64_SSE41.exe</file_name> <main_program/> </file_ref> </app_version> <app> <name>astropulse</name> </app> <file_info> <name>ap_5.00r103_SSE3.exe</name> <executable/> </file_info> <file_info> <name>libfftw3f-3-1-1a_upx.dll</name> <executable/> </file_info> <app_version> <app_name>astropulse</app_name> <version_num>500</version_num> <file_ref> <file_name>ap_5.00r103_SSE3.exe</file_name> <main_program/> </file_ref> <file_ref> <file_name>libfftw3f-3-1-1a_upx.dll</file_name> </file_ref> </app_version> </app_info> Ive just changed the app name in my app info to the above, however, does libfftw3f-3-1-1a_upx.dll have to be mentioned???? |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14654 Credit: 200,643,578 RAC: 874 |
does libfftw3f-3-1-1a_upx.dll have to be mentioned???? Yes |
Mike Davis Send message Joined: 17 May 99 Posts: 240 Credit: 5,402,361 RAC: 0 |
super :) |
Cosmic_Ocean Send message Joined: 23 Dec 00 Posts: 3027 Credit: 13,516,867 RAC: 13 |
Alright, just installed on my main cruncher. Three APs with r69 were already in progress, resume seems to play nicely. The three WUs are: - wuid=394799177 (75.110%) - wuid=395584062 (32.550%) - wuid=395743486 (12.050%) Those percentages are at the time of posting this. Not expecting anything abnormal, but the possibility is always there of course. Linux laptop: record uptime: 1511d 20h 19m (ended due to the power brick giving-up) |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
does libfftw3f-3-1-1a_upx.dll have to be mentioned???? Nope. Not unless you use Cuda or Stock application in there. (See stderr.txt text message right next to my name. FFTW is statically linked now. ) i.e. To clarify, Optimised Astropulse r103 for Windows no longer requires the fftw DLL to be present, nor reference to it in the app_info file in its application section. [Edit: But it won't hurt to leave it there either...](As long as you left the file there to match of course). "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. |
MadMaC Send message Joined: 4 Apr 01 Posts: 201 Credit: 47,158,217 RAC: 0 |
OK, I downloaded this and stopped boinc and copied the app_info and new .exe over to my projects folder. It is running well and seems faster, but.. All my CUDA workunits have disappeared, I now have only AP WU's and although in messages I have CUDA co-prcessor found, my GPU is no longer crunching, probably because there are no longer any CUDA units. I have about 50 AP units, but my cache was set to 10 days tried setting CPU to 3 in the cc_config file, but it has made no difference. How do I get BOINC to download CUDA work units again. |
Claggy Send message Joined: 5 Jul 99 Posts: 4654 Credit: 47,537,079 RAC: 4 |
OK, I downloaded this and stopped boinc and copied the app_info and new .exe over to my projects folder. By Putting entries in the app_info for setiathome Cuda as well as AstroPulse. Claggy |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
I will re-pack CUDA/opt AP package soon, after that you could use it to run both CUDA app and latest CPU AstroPulse opt app (r103 is faster, sometimes considerably faster than prev release, so it not nessesary but worth to update to it) |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
How do I get BOINC to download CUDA work units again. Right now you have full cache for SETI project. So you should crunch some AP tasks before you will be able to get new MB tasks for CUDA app. |
MadMaC Send message Joined: 4 Apr 01 Posts: 201 Credit: 47,158,217 RAC: 0 |
OK, will crunch away.. Just looking at the time for ap with the new revision - its alot faster - well done guys, your work is really appreciated amongst all us crunchers :-) |
arkayn Send message Joined: 14 May 99 Posts: 4438 Credit: 55,006,323 RAC: 0 |
|
JarrettH Send message Joined: 14 Nov 02 Posts: 97 Credit: 25,385,250 RAC: 95 |
Edit correction...figured out myself! |
John Neale Send message Joined: 16 Mar 00 Posts: 634 Credit: 7,246,513 RAC: 9 |
... r103 is faster, sometimes considerably faster than prev release, so it not nessesary but worth to update to it First off, thanks and well done to the developers of the new release of the AP optimised app. I would be interested to know from the developers what the new optimised features are, and which systems are likely to see the largest performance enhancement. |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
Revision 103 is primarily a minor performance update, with some attempts to mitigate some persistent known issues related to boincapi/AstroPulse file handling, arising from stock: - '0'd statefile' occurrences are further reduced over r69 / stock. - 'Exited with no finished file' warning likelihood is reduced. - Chance of truncated stderr.txt & result files is reduced. - annoying XML parse warnings in stderr.txt are blocked. Additionally, - FFTW library is now statically linked in, and so it is no longer required to include the DLL or app_info references to it (though if the app_info file does reference it the file must be present, but will not be used). - special handling of tasks that have reached the repeating pulse limit (30) which are prevalent it seems, now process them faster. Performance figures, depending on selected build and target platform, have been seen to range from about 1.2x to 1.5x r69 release (3x-4x stock). It should be noted that significant Berkeley stock changes appear to be in the pipeline, so the lifespan of this build may be relatively short. Possibly even from a week to a couple of months. On that note, is not considered a mandatory update, as processed results are fully compatible with those of r69 & stock version 5. Jason "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. |
Odan Send message Joined: 8 May 03 Posts: 91 Credit: 15,331,177 RAC: 0 |
Well done to all involved. I'm seeing on my first batch of 4 AP with r103 approx 1.5x performance on r69. I did have a bit of a hiccup when I installed last night - all my AP stock, both in progress and waiting to run went to computation error. Yes, I know, I should have backed up & isolated myself from the net but for the first time in ages I didn't:( As far as I could see I had done everything OK, this wasn't the first time I've updated an app, after all. Never mind - worse things happen at sea. So I would encourage anyone running r69 to try r103 but learn from my mistake. |
MarkJ Send message Joined: 17 Feb 08 Posts: 1139 Credit: 80,854,192 RAC: 5 |
It should be noted that significant Berkeley stock changes appear to be in the pipeline, so the lifespan of this build may be relatively short. Possibly even from a week to a couple of months. Hi Jason, Would that be the cuda version of AP by any chance? BOINC blog |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
It should be noted that significant Berkeley stock changes appear to be in the pipeline, so the lifespan of this build may be relatively short. Possibly even from a week to a couple of months. Hope yes. In debugging/development now. (but not connected with current Berkeley's changes in code Jason wrote about) |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
It should be noted that significant Berkeley stock changes appear to be in the pipeline, so the lifespan of this build may be relatively short. Possibly even from a week to a couple of months. The upcoming changes I was referring to in the stock application relate to additional processing of the signal in the existing non-cuda application. At this point I have no idea how much longer this would make processing, or how long before they are introduced at beta. "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. |
MadMaC Send message Joined: 4 Apr 01 Posts: 201 Credit: 47,158,217 RAC: 0 |
After borking my entire seti install (my mistake) with these last night, I will try again tonight. I dump the .exe into the projects directory - last night I did the same with the app_info as well and it all went wrong, I lost all my cuda WU's and then had problems rolling back, when I did, I was getting 'no registered app' errors as well as 'no work available' messgaaes from the server. In the end I completely uninstalled and re-installed from stock I am now running stock with the 6.6 manager and 6.08 CUDA client. Do I merge the existing app_info with the one in the R103 directory I downloaded? Just want to check so I dont bork everything again!! |
©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.