Message boards :
Number crunching :
Lunatics Windows Installer v0.41 Release Notes
Message board moderation
Previous · 1 . . . 8 · 9 · 10 · 11 · 12 · 13 · 14 . . . 17 · Next
Author | Message |
---|---|
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
Yes, this looks like the correct way. I look forward to whatever marginal updates you have to offer. I shall download them forthwith. "Freedom is just Chaos, with better lighting." Alan Dean Foster |
William Send message Joined: 14 Feb 13 Posts: 2037 Credit: 17,689,662 RAC: 0 |
hi / just a stupid beginer's request If I'm not mistaken a 650Ti is a Kepler so you want to choose the cuda 5.0 app not cuda 32. Crunch times have increased but not doubled - if you are running multiple tasks at once you should drop by at least one. Your hosts are hidden, so I can't check individual task performance and accuracy. A person who won't read has no advantage over one who can't read. (Mark Twain) |
__W__ Send message Joined: 28 Mar 09 Posts: 116 Credit: 5,943,642 RAC: 0 |
Back after some days of testing and hunting down my problems runing a complet AP-WU with the r1797-opt version. First of all, their is no problem creating/reading/writing the corresponding *.wisdom file. I observed this with ProcessMonitor(Sysinternals) and this works fine :-) Second step, i looked at the process itself with ProcessExplorer(Sysinternals). I discovered, AP6_win_x86_SSE_CPU_r1797.exe is splitt up into three threads on my Win XP Pro and because i run the CPU-Usage at 85% (for temperature reasons) the 'main' thread is set to 'Wait:Suspended' for a ~1 sec. every 5-6 sec. Sometimes this seems to interfering the other threads whitch end up in a 'Wait:Suspended' state from witch they never come back and so the proccesing of the AP-WU stopps :-( . I found tree methods to start up the stuck WU: First suspend all Seti-WUs from processing and then start them up again. How ever, works not every time. ??? Second close Boinc competly and start again. This method forces a start of chrunching from the last checkpoint and works every time. (Thanks Sten-Arne for this) Third is using the ProcessExplorer, locate the 'Wait:Suspended' thread and resume it. As far i can see this works every time and start up the WU at the position it stops :-). All methods results in a correct and validated AP-WU. My summery of this, the r1797 version don't like the option 'CPU-Usage'. I crosscecked this with a ~10 hour run on 100% and their was no hanging in sight. I am back to the 'old' r557-opt because it runs clearly on 85% CPU-Usage and the babysitting of AP-WUs costs to much time. Hopefully someone can get any benefit from my investigation. __W__ _______________________________________________________________________________ |
Mark Henderson Send message Joined: 9 Mar 02 Posts: 41 Credit: 3,964,939 RAC: 0 |
I have a Nvidia 680 and a 570 in same system. Is it ok to install Cuda 5 option to work with both? It seems to be working ok. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14649 Credit: 200,643,578 RAC: 874 |
... That's a very clear and plausible bit of analysis - and it should be easily verifiable and repeatable in testing. I'll make sure it gets back to the developer. In the meantime, you might like to experiment with Fred's TThrottle as an alternative solution for temperature management. That uses a much more fine-grained on-off cycle then BOINC crude whole seconds at a time: it would be interesting to hear what effect this has on the wait-suspended state of the other threads. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14649 Credit: 200,643,578 RAC: 874 |
I have a Nvidia 680 and a 570 in same system. Is it ok to install Cuda 5 option to work with both? It seems to be working ok. Yes. |
Fred E. Send message Joined: 22 Jul 99 Posts: 768 Credit: 24,140,697 RAC: 0 |
There was an important update to the BOINC source code yesterday, which affects all BOINC applications (across all projects, not just us) - especially GPU apps. So it's just possible that we may have to update the installer with another full release, sooner rather than later. If so, that would be an ideal opportunity to deploy whatever solution we come up with. Is that a caution to avoid new BOINC Betas until the picture clears up? I don't usually run their Betas, but some do. Kudos to BilBg for finding the cause and cure for my one line app_info.xml file, and for having the patience to show me how to confirm and fix it. I had worked around it but was concerned that some others might encounter this problem. As to why my path had only the NVIDIA string, I think I will blame it on some unknown installer - likely some of the junk my grandson has installed. I spent most of yesterday cleaning out his stuff after a spyware scare, and he's getting a good lecture today. Think if it was a NVIDIA installer, you'd have a lot more people with the problem. Another Fred Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14649 Credit: 200,643,578 RAC: 874 |
There was an important update to the BOINC source code yesterday, which affects all BOINC applications (across all projects, not just us) - especially GPU apps. So it's just possible that we may have to update the installer with another full release, sooner rather than later. If so, that would be an ideal opportunity to deploy whatever solution we come up with. No, not a problem with the BOINC Betas - well, no more than the usual quota of problems, anyway. This is a quite different section of code - the API (Application Programming Interface) - a relatively small but important section which has to be embedded in every science application, like the ones we distribute with the installer. The API handles things like updating the progress %age in BOINC Manager, and responding to BOINC's instructions to suspend and resume. Kudos to BilBg for finding the cause and cure for my one line app_info.xml file, and for having the patience to show me how to confirm and fix it. I had worked around it but was concerned that some others might encounter this problem. As to why my path had only the NVIDIA string, I think I will blame it on some unknown installer - likely some of the junk my grandson has installed. I spent most of yesterday cleaning out his stuff after a spyware scare, and he's getting a good lecture today. Think if it was a NVIDIA installer, you'd have a lot more people with the problem. Kudos indeed. |
Juha Send message Joined: 7 Mar 04 Posts: 388 Credit: 1,857,738 RAC: 0 |
That looks plausible. On an almost vanilla Windows XP with BOINC-related additions, the path is: I'm not quite sure how to read that and then there's that smiley too. Just to be clear, I wasn't accusing BOINC or Lunatics installer of anything. Did you take that from Command Prompt or from the GUI thingy? It's expanded in Command Prompt. edit: Of course Windows doesn't like the idea of moving installation directory once it's installed so the expanded vs "variable" only matters when speaking of other people's computers. OK, I've got the message - and many, many thanks for continuing to explore this while I caught my breath after Monday. OK, now we know where the problem lies, I've got two or three possibilities for a workround - I'll let them marinade for a day or two. %systemroot%\system32\find.exe would be too simple, wouldn't it? Despite the name of the directory it should work on 64-bit Windows too. The %systemroot% might be a NT thing, whereas Win 9x uses %windir%. I couldn't find any good, authoritative looking source from Microsoft's website that would list what is supported and where. |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14649 Credit: 200,643,578 RAC: 874 |
That looks plausible. On an almost vanilla Windows XP with BOINC-related additions, the path is: I was blaming NVidia at the time, but I took it from the command prompt and I wasn't aware that it might have been expanded on display - in which case, I withdraw the allegation until I can find that GUI thingy. [Edit - the GUI thingy says "C:\Program Files\NVIDIA Corporation\PhysX\Common;%SystemRoot%\system32;%SystemRoot%;%SystemRoot%\System32\Wbem;C:\CUDA\bin" - panic over. Sorry] We don't have to worry about Win 9x in this context - we have a separate and simpler installer for that (not yet updated for v7 - that will come later when and if we're convinced that the need still exists). I have WinXP thru Win7/64 hosts available for testing here, so I can try the %systemroot% approach with a deliberately broken path statement - that would indeed be the simplest solution. |
__W__ Send message Joined: 28 Mar 09 Posts: 116 Credit: 5,943,642 RAC: 0 |
That's a very clear and plausible bit of analysis - and it should be easily verifiable and repeatable in testing. I'll make sure it gets back to the developer.Thank you Richard. I know Freds TThrottle because i am using his BioncTasks, but until now there was no need to use it ;-) and for "normal" temperature monitoring i use Core Temp. If i find some time, i will have a deeper look a TThrottle over the weekend. __W__ _______________________________________________________________________________ |
BilBg Send message Joined: 27 May 07 Posts: 3720 Credit: 9,385,827 RAC: 0 |
I have WinXP thru Win7/64 hosts available for testing here, so I can try the %systemroot% approach with a deliberately broken path statement - that would indeed be the simplest solution. Of course first check: Set FindExe=%SystemRoot%\system32\find.exe IF [NOT] EXIST %FindExe% ..... for %%s in (*.aistub) do echo %%s & type %%s | %FindExe% /v "app_info" >> app_info.xml (I think there is a way/command to search a file in all Path= directories (to see if find.exe is somewhere in the Path=) but I don't remember it now) If we are paranoid and think that someone will name the Windows directory with spaces (e.g. C:\W i n d o w s) (and how Windows directory is named on German, China, etc. versions of Windows?) probably we need to use "": Set FindExe="%SystemRoot%\system32\find.exe" Â - ALF - "Find out what you don't do well ..... then don't do it!" :) Â |
Raistmer Send message Joined: 16 Jun 01 Posts: 6325 Credit: 106,370,077 RAC: 121 |
That's a very clear and plausible bit of analysis - and it should be easily verifiable and repeatable in testing. I'll make sure it gets back to the developer.Thank you Richard. TThrottle should be the best 4th path from 3 you listed earlier. It will suspend and resume app as it should. BOINC API bug makes BOINC's own method unreliable. And I'm not sure that last fix in BOINC API fixes this issue too. Their attention was on critical section mostly and AP CPU doesn't make use of critical section at all. BTW, worth to report your observations on BOINC dev or alpha mail list. SETI apps news We're not gonna fight them. We're gonna transcend them. |
Kafo Send message Joined: 17 Dec 00 Posts: 19 Credit: 15,432,367 RAC: 18 |
Attention those failing to getting the Brook app to run. I've posted a revised aistub here. Please give it a try. I have updated my app-info file with the supplied xml - then had to wait to actually get a work unit. The task has started running without trashing the work unit immediately. I will let you know if it completes successfully. Thanks for all you work |
Link Send message Joined: 18 Sep 03 Posts: 834 Credit: 1,807,369 RAC: 0 |
If i find some time, i will have a deeper look a TThrottle over the weekend. Maybe instead of a deeper look at one more workaround you should have a deeper look at the actual issue: your system cooling. Properly build computers should run 24/7@100% without any temperature issues. Just a thought. |
Mike Send message Joined: 17 Feb 01 Posts: 34253 Credit: 79,922,639 RAC: 80 |
If i find some time, i will have a deeper look a TThrottle over the weekend. Except laptops/notebooks etc. Otherwise i`d agree. With each crime and every kindness we birth our future. |
Link Send message Joined: 18 Sep 03 Posts: 834 Credit: 1,807,369 RAC: 0 |
If i find some time, i will have a deeper look a TThrottle over the weekend. Those are also pretty much never "properly build". I also had to modify the case on my laptop where the air gets in and out and set the CPU voltage properly (it was set over 300mV higher than this particular CPU needs). Result was 25°C less on the CPU and acceptable noise level. Now it crunches SETI 24/7 since years without any issues. |
Grant (SSSF) Send message Joined: 19 Aug 99 Posts: 13720 Credit: 208,696,464 RAC: 304 |
If i find some time, i will have a deeper look a TThrottle over the weekend. High end laptops should be able run run 100% load for extended periods- that's the whole point have having such a beast. But certainly the current mid-range to low end laptops, and all notebooks/ultra books are very much thermally limited. Even the highend ones though capable of considerable work are only designed to do so for very short periods of time. Being small & looking pretty comes with significant thermal limitations. Grant Darwin NT |
kittyman Send message Joined: 9 Jul 00 Posts: 51468 Credit: 1,018,363,574 RAC: 1,004 |
The kitties need help, please. I have 3 rigs throwing wobblies with the Lunatics installed GPU AP app. How can I fix this? I have made no changes to the auto installed defaults, other than changing my count to .5 in all instances in the app_info file. Here is one error task from each rig..... Host 2353446 Host 2645052 Host 3480243 Meow? "Freedom is just Chaos, with better lighting." Alan Dean Foster |
Mike Send message Joined: 17 Feb 01 Posts: 34253 Credit: 79,922,639 RAC: 80 |
Please try what i suggested already. With each crime and every kindness we birth our future. |
©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.