 |
|
Questions and Answers :
Windows :
climate prediction crashes
Message board moderation
| Author | Message |
|---|---|
 old_user35334Send message Joined: 8 Jan 05 Posts: 13 Credit: 115,849 RAC: 0 |
I am getting constant crashes of climate predictor after moving up to 4.25 and back to 4.19. I upgraded to Boinc 4.25 and restarted processing projects. Boinc crashed after a few minutes and produced client errors on all my projects. I lost all the credit for the work in prgress. I then uninstalled and went back to 4.19. seti, einstein and predictor are OK but everytime i run climate predictor the same happens. After the crash my windows/desktop become corrupted. i also get fortran error popups after the crash on occasion. This seems to be specific to climate predictor as all other Boinc projects are working fine. Before my upgrade to 4.25 I successfully completed 2 WUs each of which ran for saround 600 hours continuously with no problem. i have stress tested and benchmarked my pc - both CPU and memory. CPU temp is under 55C even at 100% load so not likely to be my HW. I have un-installed and reinstalled 4.19 but did not improve things. For now i am running the other projects but would like to process climate projects as i have a P4 3G with plenty of RAM so can contribute to the longer project quite well. Any thoughts ? |
|
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
Hi Dave You should be running OK with those specs. We'll need a bit more info, so here goes. Do you have an ATI graphics card? Some people are currently having problems with some versions of these cards. Does the crash happen when you look at the visualisation? Or consistently at some other point? Perhaps when you run some other program? Do you have the screen saver on or off? You may like to keep an eye on <a href="http://www.climateprediction.net/board/viewtopic.php?t=2822"> this</a> thread on the subject. Les |
|
old_user35334 Send message Joined: 8 Jan 05 Posts: 13 Credit: 115,849 RAC: 0 |
> Hi Dave > You should be running OK with those specs. We'll need a bit more info, so here > goes. > > Do you have an ATI graphics card? Some people are currently having problems > with some versions of these cards. > Does the crash happen when you look at the visualisation? Or consistently at > some other point? > Perhaps when you run some other program? Do you have the screen saver on or > off? > > You may like to keep an eye on <a> href="http://www.climateprediction.net/board/viewtopic.php?t=2822"> this</a> > thread on the subject. > > Les > >Hi Les, thanks for the response. I don't use the BOINC screen saver and never look at the visualisation. I use the regular windows screensaver. Crash seems to happen quite quickly after the new WU downloads- not sure that it even really starts up. Sometimes i get a popup with a memory exception and sometimes a fortran error. you can see from my work history that i compeleted a couple of units with no problem. I then did the upgrade to 4.25 - from then the client has tried to download new WUs and they all crash. I have not added any apps between WUs. In fact - i was running 4.19 - was perhaps 24hrs into a WU. suspended - did the upgrade and then the problems started. even a full uninstall/reinstall of BOINC has not helped. The crash does not seem to be associated with a particular program that I am running - it happens spontaneously as far as i can tell. I have an NVidia graphics card (ASUS 9560 with the NVidia GeForce FX5600 with 128MB of RAM and the latest drivers (7184)). My motherboard is an ASUS P4C800DX - which has the intel 875P chipset with 800MHz FSB. I have 1G of Corsair PC3700 RAM which if anything is underclocked as its good to 233MHz, not the 200MHz I am runnning. currently my system clock is 200MHz so everything is setup to spec, plus temepratures are low and stress test passes. My PC is hyperthreaded so it runs 2 models at once (seems to appear as 2 processors to BOINC)- again was not a problem until i upgraded. I get 100% CPU utilisation when it is running 2 models. I have 80G of HD free (SATA RAID pair - Promise controller on the motherboard) with 5G allocated to BOINC. I am running WinXP SP2 fully patched, with the windows firewall off (as i have a HW firewall) and AntiVir for virus protection. The only slightly unusal thing about my PC is that my boot drive is f: - does your 4.2x SW expect the main drive to be c: ? I know of one simulator (Matlab)that expects to run from c: - did you use Matlab to generate the math code ? BTW - the model that was running was hadsm3um_4.10_windows_intelx86 I notice that 4.12 is now available so i am going to try it. Dave |
|
old_user35334 Send message Joined: 8 Jan 05 Posts: 13 Credit: 115,849 RAC: 0 |
> > Hi Dave > > You should be running OK with those specs. We'll need a bit more info, so > here > > goes. > > > > Do you have an ATI graphics card? Some people are currently having > problems > > with some versions of these cards. > > Does the crash happen when you look at the visualisation? Or consistently > at > > some other point? > > Perhaps when you run some other program? Do you have the screen saver on > or > > off? > > > > You may like to keep an eye on <a> > href="http://www.climateprediction.net/board/viewtopic.php?t=2822"> > this</a> > > thread on the subject. > > > > Les > > > >Hi Les, > > thanks for the response. > I don't use the BOINC screen saver and never look at the visualisation. I use > the regular windows screensaver. Crash seems to happen quite quickly after the > new WU downloads- not sure that it even really starts up. Sometimes i get a > popup with a memory exception and sometimes a fortran error. you can see from > my work history that i compeleted a couple of units with no problem. I then > did the upgrade to 4.25 - from then the client has tried to download new WUs > and they all crash. I have not added any apps between WUs. In fact - i was > running 4.19 - was perhaps 24hrs into a WU. suspended - did the upgrade and > then the problems started. even a full uninstall/reinstall of BOINC has not > helped. The crash does not seem to be associated with a particular program > that I am running - it happens spontaneously as far as i can tell. > > I have an NVidia graphics card (ASUS 9560 with the NVidia GeForce FX5600 with > 128MB of RAM and the latest drivers (7184)). My motherboard is an ASUS > P4C800DX - which has the intel 875P chipset with 800MHz FSB. I have 1G of > Corsair PC3700 RAM which if anything is underclocked as its good to 233MHz, > not the 200MHz I am runnning. currently my system clock is 200MHz so > everything is setup to spec, plus temepratures are low and stress test passes. > My PC is hyperthreaded so it runs 2 models at once (seems to appear as 2 > processors to BOINC)- again was not a problem until i upgraded. I get 100% > CPU utilisation when it is running 2 models. > > I have 80G of HD free (SATA RAID pair - Promise controller on the motherboard) > with 5G allocated to BOINC. I am running WinXP SP2 fully patched, with the > windows firewall off (as i have a HW firewall) and AntiVir for virus > protection. > > The only slightly unusal thing about my PC is that my boot drive is f: - does > your 4.2x SW expect the main drive to be c: ? I know of one simulator > (Matlab)that expects to run from c: - did you use Matlab to generate the math > code ? > > BTW - the model that was running was hadsm3um_4.10_windows_intelx86 I notice > that 4.12 is now available so i am going to try it. > > Dave > > update 4.12 crashes as well - totally corrupts my windows envrinoment. fonts in web browser. i ran with no anti virus so its not an interaction. i'm going to uninstall BOINC again and stop running cpdn for a few months. |
|
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
Dave I'm just a user, so I don't know about Matlab, although I think I have seen it advertised in New Scientist. Boot drive = f: That is the only thing in your setup that is unusual. Could be the problem, as hadsm is a very complex program, and who knows what parts of the code may do. It was developed over 20-30 years by numerous scientists/programmers working for the Met Office. <a href="http://www.metoffice.com/corporate/pressoffice/2005/pr20050412.html"> This</a> is the latest machine to be running it. And various other Met programs. Some computers have been unable to run CP. The reasons have never been found, as the owners have given up. Two more notes: I've just completed a second model using 4.10, and am now running a second 4.12 model. The first is just past 230 hours, (32.56%), and the second is just short of 17 hours. and 2) one doesn't pick which hadsm version to run. You get whatever is available from the pool of parameter sets. Les |
|
old_user35334 Send message Joined: 8 Jan 05 Posts: 13 Credit: 115,849 RAC: 0 |
> Dave > I'm just a user, so I don't know about Matlab, although I think I have seen it > advertised in New Scientist. > > Boot drive = f: > That is the only thing in your setup that is unusual. Could be the problem, as > hadsm is a very complex program, > and who knows what parts of the code may do. It was developed over 20-30 years > by numerous scientists/programmers > working for the Met Office. <a> href="http://www.metoffice.com/corporate/pressoffice/2005/pr20050412.html"> > This</a> is the latest machine to be running it. And various other Met > programs. > > Some computers have been unable to run CP. The reasons have never been found, > as the owners have given up. > > Two more notes: > I've just completed a second model using 4.10, and am now running a second > 4.12 model. The first is just past > 230 hours, (32.56%), and the second is just short of 17 hours. > and 2) > one doesn't pick which hadsm version to run. You get whatever is available > from the pool of parameter sets. > > Les > >Les - Thanks for the explanation on the model- i did not realise that it was a random selection from the pool - the only solace i have is that until I upgraded to BOINC 4.25 I was happily running CPDN models. So even though my boot drive is f: previous CPDN SW was fine. Maybe the new model exercises some different older code. Oh well - I'm probably going to give up on cpdn for a while. The crashes mess up my desktop, erase buttons on my browser and erase WU from other projects so I can't really live with them. |
|
Send message Joined: 18 Nov 04 Posts: 2 Credit: 323,950 RAC: 0 |
I'm just going to interrupt you to tell that I have been experiencing exactly the same porblems on my machine. My machine went from 100% stable at everything, to crashing several times a day. It didn't actually crash, but graphics gets messed up, network connection is disturbed (cant connect to other PC's in the network, but can browse the internet with Firefox) and programs report out off memory errors when opening them. Basically the PC is unusable. One time it actually messed up my hard drive so bad, Winfdows couldn't boot, had to reformat and reinstall :-( The PC is an Athlon Sempron 2800+ (2GHz) on an ASUS A7V880 with 1Gb of dual channel memory. Graphics card is an ATI Radeon 9500Pro. I don't use the screensaver. This problem is extremely annoying, and I have had to Detach from the project. |
|
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
Karsten Welcome to the forum. There are several problems at the moment. 1) The trickle server is down. 2) hadsm has become unusable with Win98; which doesn't affect you. We hope the programmer can do something about this. Someday. 3) Some versions of BOINC are not working with some versions of ATI cards. See <a href="http://setiweb.ssl.berkeley.edu/forum_thread.php?id=12948"> this</a> thread at SETI. It MAY affect you. And some people, such as Dave Park, are having lots of problems with which we don't seem to be able to help. Some of the problems seem to have started when hadsm was re-compiled to be 'signed', which is now required by the latest versions of BOINC. Some people have got going again by reverting to BOINC 4.19 Have you run the stability test programs mentioned <a href="http://www.climateprediction.net/board/viewtopic.php?t=2126"> here?</a> Les |
|
old_user35334 Send message Joined: 8 Jan 05 Posts: 13 Credit: 115,849 RAC: 0 |
> I'm just going to interrupt you to tell that I have been experiencing exactly > the same porblems on my machine. > > My machine went from 100% stable at everything, to crashing several times a > day. It didn't actually crash, but graphics gets messed up, network connection > is disturbed (cant connect to other PC's in the network, but can browse the > internet with Firefox) and programs report out off memory errors when opening > them. Basically the PC is unusable. > > One time it actually messed up my hard drive so bad, Winfdows couldn't boot, > had to reformat and reinstall :-( > > The PC is an Athlon Sempron 2800+ (2GHz) on an ASUS A7V880 with 1Gb of dual > channel memory. Graphics card is an ATI Radeon 9500Pro. > > I don't use the screensaver. > > This problem is extremely annoying, and I have had to Detach from the > project. > > >Karsten - I just found that interaction between my anti-virus SW and the later models seem to have been the cause of my crashes. I uninstalled the SW I use, and now the constant crashes have stopped - I am sending trickles (although the server is slow right now). I was using antivir version 6 with winxp home SP2 - both at the latest patch levels. Try a full unistall of your antivirus and let everyone know what you were using and if it fixed it. Also one of my PC's had a problem with zone alarm interaction causing crashes as well. Checkout this thread for a lot of data on other people's experiences with the latest models. http://www.climateprediction.net/board/viewtopic.php?t=2822 Dave |
|
Send message Joined: 18 Nov 04 Posts: 2 Credit: 323,950 RAC: 0 |
Hey again. Sorry for the long time to answer, but real life has taken much of the time I could spend in front of my computer, so I have not had time to respond untill today. Les Bayliss: The problem with the ATI-cards seems to be about displaying the graphics, this works on my system, but I normaly don't use it. I have run Memtest86 through several tests, all passed OK. SuperPI to 32M iterations (takes forever) passes. Prime95's torture has been run overnight several times, witout errors of any kind. The computer behaves ansolutely perfect at anything I throw at it, but letting hadsm execute brings it to a semi crashed state before 2% is finished. Dave Park: I am actually running Antivir version 6 on a WinXP SP2, also uptodate with windows updates. I have not tried stopping the antivirus program, since I would then be defenceless against viruses, and I don't want to do that. But the fact that we are both experiencing the same problems on the same OS with the same version of AV, should help the developers fix hadsm. I put the blame on hadsm, since it is the only progam I have experienced to have an incompability with Antivirus. For now my conclusion is that the problem is probably caused by an incompatability with Antivir version 6 together with WinXP SP2. As not having antivirus running on my system is not an option, I will have to not run climateprediction on this machine until the problem is reported to be solved. This leaves climateprediction running 15% of the time on my other machine, a Celeron 1800 which has a turnaround time of about 80 days. Thanks for your replies, allways nice that peoble are ready to help, and to know that I'm not the only one with this problem. |
©2024 climateprediction.net
