Questions and Answers :
GPU applications :
workunit times & flops question
Message board moderation
Author | Message |
---|---|
rwarren873 Send message Joined: 2 Sep 06 Posts: 25 Credit: 39,741,063 RAC: 87 |
I've tried to puzzle out how to fix this but it's apparently beyond me. I've read the few threads about it I can find here but I just don't understand them, so I apologize in advance.. I've been running Lunatics .40 for a while now and everything seems to be going fine, except Boinc grossly over-estimates astropulse_v6 6.01 GPU (ATI)and setiathome 6.10 task completion times. Overestimating like 200ish hours for a AP task and 40 hours for a setiathome_enhanced 6.10 task. I believe that's why I get so few of those tasks. This is one of the tasks in my event log (I don't have an AP task at the moments so I can't post one of them): 2/19/2013 6:52:10 PM | SETI@home | Starting task 29dc12af.26096.4566.10.10.40_0 using setiathome_enhanced version 610 (ati13ati) in slot 14 I've worked out that I *can* fix it by adding the <flops> tag in my app_info file and that I'm supposed to get the flops for my GPU from my computer details page here: http://setiathome.berkeley.edu/host_app_versions.php?hostid=6585335 The big problem is I have a honking big app_info file, and I'm not quite sure WHICH of all the entries I should be adding a <flops> tag to. I've also read that Boinc is supposed to adjust the times, but may not due to using app_info. Other posts say to avoid using <flops> to avoid deadline issues if the task type changes for whatever reason. I'm currently using Boinc 7.28 if that matters. I'll post my app_info if that'd help too. Any help would be appreciated! |
rwarren873 Send message Joined: 2 Sep 06 Posts: 25 Credit: 39,741,063 RAC: 87 |
I guess I should add that I've modified app_info to run two concurrent GPU workunits. The 6.10 tasks take about 40 minutes and the AP tasks can take as much as 7 hours. Boinc also occasionally runs one SETI and one Einstein GPU workunit concurrently, but that shouldn't affect the processing speed numbers from my computer application details page - as I understand it anyway. Einstein workunits take 1:50 consistently. |
skildude Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 |
just a reminder that the GPU apps for AP and MB WU's have been updated since the unified installer was put out. http://lunatics.kwsn.net/index.php?module=Downloads;sa=dlview;id=355 http://lunatics.kwsn.net/index.php?module=Downloads;sa=dlview;id=357 In a rich man's house there is no place to spit but his face. Diogenes Of Sinope |
rwarren873 Send message Joined: 2 Sep 06 Posts: 25 Credit: 39,741,063 RAC: 87 |
just a reminder that the GPU apps for AP and MB WU's have been updated since the unified installer was put out. It appears that doing those updates is more complicated than replacing the filenames in my app_info. Nothing but BSOD until I changed them back. Would they have solved the run-time estimate problem I described anyway? At any rate, those updates will have to wait for them to sort out the legal issues for the installer - or for me to have WAY more time to figure out what's the problem in app_info. |
skildude Send message Joined: 4 Oct 00 Posts: 9541 Credit: 50,759,529 RAC: 60 |
you'll need to install the appropriate apps and additional files into your seti@home folder. then rename the apps in your app_info. restart BOINC and it should work fine In a rich man's house there is no place to spit but his face. Diogenes Of Sinope |
rwarren873 Send message Joined: 2 Sep 06 Posts: 25 Credit: 39,741,063 RAC: 87 |
I did that, but the runtime or command line parameters must need to be changed too. Are you sure no tinkering with parameters is needed? I changed nothing but the file names and as soon as it started a GPU app, I got the BSOD. |
©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.