Message boards :
Graphics cards (GPUs) :
A lot of Failures on XP with a GTX470
Message board moderation
| Author | Message |
|---|---|
|
Send message Joined: 15 Oct 12 Posts: 18 Credit: 27,916,718 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
I am getting a lot of errors after "upgrading" to a GTX470 on a dedicated cruncher. I was running a GTX275 before. I know the 470 is still an older card, but the price was right. Free, thanks to a team-mate! Anyway, I am getting 2 different errors: Exit status 80 (0x50) Unknown error number -97 (0xffffffffffffff9f) Unknown error number Also, I noticed in the log that BOINC is suspending and resuming a LOT even on tasks that complete normally. I have the TThrottle running and keeping the GPU temp at or below 90 degrees Celsius. I am sure it probably throttling it back occasionally. Without TThrottle running the temp gets to around 92 degrees Celsius. I am currently running the latest version of BOINC and the latest graphics driver. I am running XP on that machine currently, but will be upgrading soon. Any ideas? Should I stop Tthrottle or set the temperature a little higher? Thanks, Chuck |
|
Send message Joined: 15 Oct 12 Posts: 18 Credit: 27,916,718 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
I also found these error messages: The simulation has become unstable. Terminating to avoid lock-up (1) SWAN : FATAL : Cuda driver error 719 in file 'swanlibnv2.cpp' in line 1963. Thanks again, Chuck |
|
Send message Joined: 16 Mar 11 Posts: 509 Credit: 179,005,236 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
I am running XP on that machine currently, but will be upgrading soon. Correct me if I am wrong but I think the general consensus amongst Windows users who crunch GPUgrid is that with respect to running GPUgrid tasks, getting rid of XP and installing Vista/7/8 is actually a downgrade in terms of both performance and stability. The only OS option available to you that is truly an upgrade is to switch to Linux... best performance, most user friendly and most stable. BOINC <<--- credit whores, pedants, alien hunters |
|
Send message Joined: 22 Nov 12 Posts: 72 Credit: 14,040,706,346 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
I am running XP on that machine currently, but will be upgrading soon. Second that! skgiven discusses this (XP vs vista/7/8)performance issue on another thread here: http://www.gpugrid.net/forum_thread.php?id=3615&nowrap=true#34939 |
|
Send message Joined: 15 Oct 12 Posts: 18 Credit: 27,916,718 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
I've been debating switching over to Ubuntu on that machine. It will be a learning curve, but you gotta learn sometime! Thanks, Chuck |
|
Send message Joined: 16 Mar 11 Posts: 509 Credit: 179,005,236 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
There is a bit of a learning curve and there is the possibility that there is something wrong with your hardware. Ubuntu can't fix a hardware problem so to avoid disappointment and a scenario like "darn, I've invested 2 months into learning Linux and my card still gets errors!", I would do a little more research into what's going wrong in your situation. Let me explain. XP uses a different driver model than Vista/7/8. As far as I know you shouldn't be getting the errors you're getting with XP, they seem to be more of a Vista/7/8 phenomenon. My advice is that if you can't get that card to run stable under XP then something else is wrong and Ubuntu probably isn't going to fix that. The card itself might be defective, for example. I would wait for someone more experienced with Windows than I am to comment. Also, I have no experience with the 4xx series of cards as my first was a 5xx. Again, one of the more experienced crunchers might have just the tip you need to get your 4xx card running right on XP alone. To me your card seems to be running rather hot. Maybe the temps you've mentioned are normal for 4xx cards but if not then one of the first things you should do (if you haven't done so already) is blow the dust out of the fan and cooling fins and, just as important, renew the thermal grease under the heat sink. If that doesn't cure the heat problem then perhaps you need to look at increasing the GPU fan speed or other measures. If you decide to install Ubuntu then be aware that AFAIK, Linux doesn't have the equivalent of TThrottle so don't depend on that as part of the solution under Linux. Anyway, throttling the GPU is probably not the best approach and it doesn't seem to be working under XP so again, there seems to be something else wrong. A better way to get the temp down would be to increase the fan speed and/or decrease the clock speeds. That's all doable on Linux and it's doable on XP too. BOINC <<--- credit whores, pedants, alien hunters |
|
Send message Joined: 2 Jan 09 Posts: 303 Credit: 7,321,800,090 RAC: 270 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
There is a bit of a learning curve. I WISH someone would come up with a package for linux that would include everything one needs, and it auto installs! That would make it ALOT easier for us Windows folks to switch over to Linux. I don't care which version of Linux it works on as long as it is a free one, but being able to have essentially a plug and play setup would make things REALLY nice and easy. Maybe even two packages, or one with the option to use or not use a gpu, and include all the lib etc files one needs to crunch for any project. I know I am dreaming again, but hey it is fun!! |
|
Send message Joined: 16 Mar 11 Posts: 509 Credit: 179,005,236 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Mikey, I've been thinking exactly the same thing. It's doable but would it be accepted and adopted? The idea deserves a new thread as it's kind of off-topic in this one. I have to go now but if you want to start a new thread on it I can join later with some ideas I have. Let's get some ideas regarding exactly what people want and what they would be willing to try and/or adopt and see what can be done. BOINC <<--- credit whores, pedants, alien hunters |
|
Send message Joined: 2 Jan 09 Posts: 303 Credit: 7,321,800,090 RAC: 270 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Mikey, I've been thinking exactly the same thing. It's doable but would it be accepted and adopted? The idea deserves a new thread as it's kind of off-topic in this one. I have to go now but if you want to start a new thread on it I can join later with some ideas I have. Let's get some ideas regarding exactly what people want and what they would be willing to try and/or adopt and see what can be done. That works for me. |
Retvari ZoltanSend message Joined: 20 Jan 09 Posts: 2380 Credit: 16,897,957,044 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
I am getting a lot of errors after "upgrading" to a GTX470 on a dedicated cruncher.... It could cause the errors in itself. You should set the BOINC manager to use the GPU while your computer is in use, and if there are some applications which need a lot of GPU power then you should put them on the exclusive application list in the BOINC manager, or suspend GPU activity in the BOINC manager manually (by right clicking on its icon) when you are using those apps. You should also set the BOINC manager to use the CPU at 100% of the time (to avoid suspending and resuming a LOT). I have the TThrottle running and keeping the GPU temp at or below 90 degrees Celsius. I am sure it probably throttling it back occasionally. Without TThrottle running the temp gets to around 92 degrees Celsius. Maybe TThrottle suspending the apps that often, so you should find another way to reduce the GPU temperature, as 90ºC is way too high. You should: 1. clean the fins of the GPU's cooler with an air duster. 2. Increase the airflow in your computer's case. (add 1 fan at a time as adding too many fans could be counter productive, or remove the side cover) 3. Increase the GPU's fan speed with a 3rd party utility like MSI Afterburner (it works with other manufacturer's cards too) 4. lower your GPU's frequency and its operating voltage. I am currently running the latest version of BOINC and the latest graphics driver. I am running XP on that machine currently, but will be upgrading soon. XP is the best Windows OS for GPU crunching. Any ideas? Should I stop Tthrottle or set the temperature a little higher? Try to keep the card's temperature as low as possible. Don't let it go over 80ºC in the long term. |
dskagcommunitySend message Joined: 28 Apr 11 Posts: 463 Credit: 958,266,958 RAC: 34 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Is that a typo to recommend to lower the voltage? ;) DSKAG Austria Research Team: http://www.research.dskag.at
|
|
Send message Joined: 15 Oct 12 Posts: 18 Credit: 27,916,718 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Zoltan - I have read a lot about that cards temperatures and I am hearing the card is fine in the low 90's and it maxes at 104... It does seem VERY high though. However, the fan on the card is barely running and when it does get to 90 it self regulates and never gets over 92. I had set Tthrottle at 90, but changed it to 92 and now I am not seeing the suspend/resume cycle anymore. The machine is a dedicated machine that sits in my VERY cold garage right now. I have to use TeamViewer to log into it remotely occasionally. So, I have the CPU and memory settings at 85% so I can log in! Thanks for the tip about MSI Afternurner. I didn't know it would work with other cards. Thanks, Chuck |
|
Send message Joined: 15 Oct 12 Posts: 18 Credit: 27,916,718 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Dagorath - Thanks for the advice. I agree that learning Linux is going to be a learning curve and that there could still be a hardware issue with that card... I am going to have to do something by the end of April anyway since XP is going end-of-life then. In the mean time, I will explore the other options you mentioned. Thanks, Chuck |
|
Send message Joined: 16 Mar 11 Posts: 509 Credit: 179,005,236 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
However, the fan on the card is barely running and when it does get to 90 it self regulates and never gets over 92. I had set Tthrottle at 90, but changed it to 92 and now I am not seeing the suspend/resume cycle anymore. That's the problem... the fan isn't ramping up fast enough. 2 of my 3 cards are like that too and I have to "manually" force the fan to a higher speed. You can use MSI Afterburner to do that on your card. If it's in a cold garage you should easily be able to keep the temp below 70*C. Though the specs may say low 90s is fine I would try for at least less than 80*C. If that all works out then I would say you won't be disappointed migrating to Linux. However, since that machine is a dedicated cruncher, I have to wonder if it makes sense to migrate. I mean yes XP is reaching end of life but if the only thing that machine does is crunch then it probably doesn't need constant security updates. And it doesn't suffer from all the things that afflict Vista/7/8. I'm not trying to talk you out of Linux, just saying don't do work you don't need to do. On the other hand, just wanting to see what Linux is all about is a good enough reason and I doubt you'll find the learning curve as nasty as a lot of people think it is. Also, Linux has some nice networking functionality built in (or installable for free) that you might find handy for managing your rigs remotely. BOINC <<--- credit whores, pedants, alien hunters |
Retvari ZoltanSend message Joined: 20 Jan 09 Posts: 2380 Credit: 16,897,957,044 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Is that a typo to recommend to lower the voltage? ;) No :) When the GPU temperature is that high - and there's no other way to reduce that - you should underclock your GPU, and as a consequence of the lower clock you can reduce the voltage also, which will further decrease the power draw of the GPU. |
|
Send message Joined: 4 Apr 09 Posts: 450 Credit: 539,316,349 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
I'm on the 'fix the fan curve' team. Because you are in a cold garage it should be easy to get the fan curve fixed to keep temps lower and you could then also stop running T-Throttle. The old 4XX GPUS are known to be power hogs and more specifically the pattern moves from linear to logarithmic after 75C. Crank the fan and let your 470 do it's best!!! Thanks - Steve |
|
Send message Joined: 15 Oct 12 Posts: 18 Credit: 27,916,718 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Everyone - I installed the MSI Afterburner. Wow what a difference! I adjusted the fan curve and my temps went from around 90 to around 70! It's in the garage now and it isn't bothering anyone. This summer when I bring it back into the utility room may be a little different, but for now it is really cool. Thanks for all the input! I may look into the heat shield grease when I bring it back inside. Chuck |
|
Send message Joined: 15 Oct 12 Posts: 18 Credit: 27,916,718 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
If that all works out then I would say you won't be disappointed migrating to Linux. However, since that machine is a dedicated cruncher, I have to wonder if it makes sense to migrate. I mean yes XP is reaching end of life but if the only thing that machine does is crunch then it probably doesn't need constant security updates. And it doesn't suffer from all the things that afflict Vista/7/8. I'm not trying to talk you out of Linux, just saying don't do work you don't need to do. On the other hand, just wanting to see what Linux is all about is a good enough reason and I doubt you'll find the learning curve as nasty as a lot of people think it is. Also, Linux has some nice networking functionality built in (or installable for free) that you might find handy for managing your rigs remotely. Good point, the only thing it does is connect to BOINC and TeamViewer. As long as I don't surf on it I should be ok with AVG running on it. Thanks for all the advice, Chuck |
|
Send message Joined: 15 Oct 12 Posts: 18 Credit: 27,916,718 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Because you are in a cold garage it should be easy to get the fan curve fixed to keep temps lower and you could then also stop running T-Throttle. So, I can stop running Tthrottle now? I guess that makes sense now that Afterburner is consistently keeping it around 70. Thanks, Chuck |
|
Send message Joined: 16 Mar 11 Posts: 509 Credit: 179,005,236 RAC: 0 Level ![]() Scientific publications ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]()
|
Because you are in a cold garage it should be easy to get the fan curve fixed to keep temps lower and you could then also stop running T-Throttle. Glad to hear you've got it down to ~70. Yes, stop running TThrottle. Are the fans running at maximum speed? I believe maximum is 80%? I'm asking because as the weather becomes warmer you'll need to increase the fan speed to maintain the 70*C target temperature. It would be nice if you had software that could monitor the actual temp and adjust the fan speed to maintain the target temp. Does MSI Afterburner or anything for Windows do that? The alternative to temperature monitoring software is to modify the firmware in the GPU's BIOS which is what Snow Crash refers to as "recurving the fan speed". BOINC <<--- credit whores, pedants, alien hunters |
©2025 Universitat Pompeu Fabra