Message boards :
Number crunching :
.vlar WUs to NVIDIA GPUs (Problem Solved)
Message board moderation
Previous · 1 · 2 · 3 · 4 · 5 · Next
Author | Message |
---|---|
Link Send message Joined: 18 Sep 03 Posts: 834 Credit: 1,807,369 RAC: 0 |
Hmmm. My recipe worked very nicely while I only had a few VLARs, and they'd all arrived in a neat contiguous block. But now I've got a boatload more, and they're all dotted around individually in ones and twos. 1. If you don't have it yet, download and install Notepad++, install it with plugins (specially TextFX). 2. Use the replace feature in regular expression mode, as search string use <result>\r\n<name>.{20,35}vlar_.</name>\r\n<final_cpu_time>0.000000</final_cpu_time>\r\n<final_elapsed_time>0.000000</final_elapsed_time>\r\n<exit_status>0</exit_status>\r\n<state>2</state>\r\n<platform>windows_intelx86</platform>\r\n<version_num>610</version_num>\r\n<plan_class>cuda_fermi</plan_class>\r\n<wu_name>.{20,35}</wu_name>\r\n<report.{30,40}deadline>\r\n<received.{30,40}time>\r\n<file_ref>\r\n<file.{40,60}name>\r\n<open.{15,30}name>\r\n</file_ref>\r\n</result>\r\n replace with nothing. Use that carefully, I didn't test it very much, it finds your example and I hope the wildcards are OK to work with all possible WU names... it *should* work, but consider it alpha. If you want some more information about the regular expresions, you can find it here. Or if something doesn't work as it should, just ask and post it together with the result, where it didn't work. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Thanks. Didn't work exactly as posted, because the xml file is indented (four space characters per tag level), but brute force soon cured that. Anyone know the RegExp for 'arbitrary white space'? (like M$ Word's ^w) |
Horacio Send message Joined: 14 Jan 00 Posts: 536 Credit: 75,967,266 RAC: 0 |
Thanks. Didn't work exactly as posted, because the xml file is indented (four space characters per tag level), but brute force soon cured that. Anyone know the RegExp for 'arbitrary white space'? (like M$ Word's ^w) \s -> Match any whitespace character (more on http://www.codeproject.com/Articles/9099/The-30-Minute-Regex-Tutorial) |
Link Send message Joined: 18 Sep 03 Posts: 834 Credit: 1,807,369 RAC: 0 |
Thanks. Didn't work exactly as posted, because the xml file is indented (four space characters per tag level), but brute force soon cured that. Anyone know the RegExp for 'arbitrary white space'? (like M$ Word's ^w) Well, use either \s{4} for to replace those 4 spaces or simply 4 "normal" spaces (which make it IMO better readable in case you need to adjust something). There are two lines with 8 spaces in it according to my client_state, there you would use \s{8}. So the proper search string would be: <result>\r\n <name>.{20,35}vlar_.</name>\r\n <final_cpu_time>0.000000</final_cpu_time>\r\n <final_elapsed_time>0.000000</final_elapsed_time>\r\n <exit_status>0</exit_status>\r\n <state>2</state>\r\n <platform>windows_intelx86</platform>\r\n <version_num>610</version_num>\r\n <plan_class>cuda_fermi</plan_class>\r\n <wu_name>.{20,35}</wu_name>\r\n <report.{30,40}deadline>\r\n <received.{30,40}time>\r\n <file_ref>\r\n <file.{40,60}name>\r\n <open.{15,30}name>\r\n </file_ref>\r\n</result>\r\n But that's more just something to start with, I don't know if that's something only for CPUs, but I have for example a fpops_cumulative tag (and of course no plan_class and windows_x86_64 as platform). So others might have some other tags or values in there as well. |
Horacio Send message Joined: 14 Jan 00 Posts: 536 Credit: 75,967,266 RAC: 0 |
I noticed that Eric (or someone) added a "Use ATI GPU" preference to the project preference page, probably for the new AP - for - ATI application. I don't have an ATI GPU, but it was set to on. I turned it off and haven't got a VLAR for Nvidia on the last 5 successful gpu work requests. Ive changed the settings to disable ATI GPUs and Ive not received any more vlars for GPUs... (Sadly, Ive not received vlars neither for CPU, so I cant know if the change made the trick or it is just that there are no vlars available to send...) |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Thanks. Didn't work exactly as posted, because the xml file is indented (four space characters per tag level), but brute force soon cured that. Anyone know the RegExp for 'arbitrary white space'? (like M$ Word's ^w) Well, the "four literal typed spaces" was the brute force approach that confirmed that it was the indenting which caused the problem. Looking at Horacio's tutorial, example (20) suggests \s* might do the trick. I'll work on it some more in the morning - I think I have enough non-VLAR CUDA tasks to last me until then. |
Link Send message Joined: 18 Sep 03 Posts: 834 Credit: 1,807,369 RAC: 0 |
Looking at Horacio's tutorial, example (20) suggests \s* might do the trick. I'll work on it some more in the morning - I think I have enough non-VLAR CUDA tasks to last me until then. Yeah, that seems to work too and you don't need to care anymore, how many spaces there are. |
S@NL - John van Gorsel Send message Joined: 5 Jul 99 Posts: 193 Credit: 139,673,078 RAC: 0 |
My GTX5xx Fermi cards seem to handle the vlar's fairly well at around 3x the time they need for a midrange task. My Linux pc with a GTX260 ends all vlar tasks after 1h50m with a -177 error (estimated processing time is now at 13 minutes). I suspended all 19 GPU vlar tasks on that pc for the time being. Seti@Netherlands website |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Looking at Horacio's tutorial, example (20) suggests \s* might do the trick. I'll work on it some more in the morning - I think I have enough non-VLAR CUDA tasks to last me until then. Yes, it worked a treat here too (after a suitable safety margin of coffee had been administered). Many thanks for the help. We don't know for certain whether the scheduler has been re-fixed to prevent VLARs being sent to NVidia cards again, unless Sutaru has had any more emails from staff? But I've not seen any new ones on any of my hosts since yesterday afternoon. |
Sutaru Tsureku Send message Joined: 6 Apr 07 Posts: 7105 Credit: 147,663,825 RAC: 5 |
As far as I know, Eric know the problem and will fix it. I'll ask him if he could send me a message if he fixed it. If, then I'll post a message here. * Best regards! :-) * Sutaru Tsureku, team seti.international founder. * Optimize your PC for higher RAC. * SETI@home needs your help. * |
Fred J. Verster Send message Joined: 21 Apr 04 Posts: 3252 Credit: 31,903,643 RAC: 0 |
Looking at Horacio's tutorial, example (20) suggests \s* might do the trick. I'll work on it some more in the morning - I think I have enough non-VLAR CUDA tasks to last me until then. I haven't seen VLARs (610) for ATI GPUs lately, only a lot of ~0.4 AR WUs. Just got a few VLARs on the GPUs. As a result, AMD/ATI GPU aren't also getting vlars? OFF TOPIC And a difference between BOINC 6.12.34 and BOINC 7.0.28/31, it reports twice the Peak FLOPS on GPUs, atleast on my HD5870, 5440GFLOPS per GPU, was 2720GFLOPS per GPU! |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13736 Credit: 208,696,464 RAC: 304 |
Got a couple more VLARs on my Nvidia GPUs overnight. Grant Darwin NT |
Sutaru Tsureku Send message Joined: 6 Apr 07 Posts: 7105 Credit: 147,663,825 RAC: 5 |
I got EMail from Eric. He fixed the problem last night (local time). So we shouldn't get .vlar WUs for NVIDIA GPUs since Thursday Aug/09 around 06:00 UTC/GMT (if I calculated correct ;-) - I guess. If your BOINC got nevertheless .vlar WUs for your NVIDIA GPU (after this above mentioned time), please drop a message here. * Best regards! :-) * Sutaru Tsureku, team seti.international founder. * Optimize your PC for higher RAC. * SETI@home needs your help. * |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
I just found 28jl10ab.20418.294615.5.10.228.vlar on one of mine. I've just had it resent to CPU, but the wingmate is still showing the original sent time of 11:09 UTC today, and the WU demonstrates that they are (or were then) still being sent to both stock and anonymous platform hosts. |
Sutaru Tsureku Send message Joined: 6 Apr 07 Posts: 7105 Credit: 147,663,825 RAC: 5 |
Eric was informed, he will look why this still happen. * Best regards! :-) * Sutaru Tsureku, team seti.international founder. * Optimize your PC for higher RAC. * SETI@home needs your help. * |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14650 Credit: 200,643,578 RAC: 874 |
Eric put on a new scheduler at 21:55 UTC. Don't worry about any VLARs sent before that time - deal with them as you wish. But please let us know about any VLARs allocated after that time to NVidia applications. |
.clair. Send message Joined: 4 Nov 04 Posts: 1300 Credit: 55,390,408 RAC: 69 |
@Fred I haven't seen VLARs (610) for ATI GPUs lately, only a lot of ~0.4 AR WUs. I am still getting VLARs on ATI, I think that part is working as normal, i dont think there are many to have today :( 09/08/2012 21:21:08 {utc} Finished download of 21fe12ac.8763.24647.8.10.168.vlar |
Sutaru Tsureku Send message Joined: 6 Apr 07 Posts: 7105 Credit: 147,663,825 RAC: 5 |
:-( 10 Aug 2012 01:06:56 UTC - .vlar WU to NVIDIA GPU. app_info.xml file with 608_cuda entry. [EDIT: I sent EMail to the admins.] * Best regards! :-) * Sutaru Tsureku, team seti.international founder. * Optimize your PC for higher RAC. * SETI@home needs your help. * |
Sutaru Tsureku Send message Joined: 6 Apr 07 Posts: 7105 Credit: 147,663,825 RAC: 5 |
@Fred .vlar WUs are OK for CPUs and AMD/ATI GPUs. NVIDIA GPUs have problems with this kind of WUs. * Best regards! :-) * Sutaru Tsureku, team seti.international founder. * Optimize your PC for higher RAC. * SETI@home needs your help. * |
S@NL - John van Gorsel Send message Joined: 5 Jul 99 Posts: 193 Credit: 139,673,078 RAC: 0 |
I didn't check all my pc's but this one is now downloading 5 vlar tasks for the Fermi_GPU. The request for tasks is from 9:43 CET. |
©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.