Message boards : Graphics cards (GPUs) : Is the problem whit GTX 260, Ok now??
Author | Message |
---|---|
Is the problem whit GTX 260, OK now?? And if that the case, what drivers and boinc version is recomanded?? | |
ID: 13091 | Rating: 0 | rate: / Reply Quote | |
Is the problem whit GTX 260, OK now?? And if that the case, what drivers and boinc version is recomanded?? In a word - No. Your best bet for the GTX260's is use them on another project until nVidia can fix things or the project comes up with a work around. If you do want to use them for GPUgrid expect around a 50% (or worse) error rate, depending on the mix of work units that you get. The last working set of drivers (for the GTX260) was the 182.50, but they only support cuda 2.1. The current app needs at least cuda 2.2 so that means the 185.xx drivers, or later are required. The project is currently recommending BOINC 6.10.3, however I have found 6.10.7 .11 and now .13 seems to be stable from my testing. ____________ BOINC blog | |
ID: 13097 | Rating: 0 | rate: / Reply Quote | |
We are implementing a workaround, which is quite drastic. We will make an application just for 260 which does not use PME. Clearly only some workunits can run there, but at least we move on without waiting. | |
ID: 13099 | Rating: 0 | rate: / Reply Quote | |
Please have it as an option not as default. Not all is affected by this bug. | |
ID: 13103 | Rating: 0 | rate: / Reply Quote | |
Please have it as an option not as default. Not all is affected by this bug. Agreed, I have been running my GTX260s for months and I don't think I have seen this problem at all, or at most once or twice in all that time. Though I am now running another project in parallel so my GPU Grid run time on those cards is reduced, I am still not that avid to lose more compute time if it is not needed... Of course, given no choice I can always select other projects for the GTX260 cards and move on ... though I would rather not ... Options are a good thing ... :) | |
ID: 13104 | Rating: 0 | rate: / Reply Quote | |
It would not matter if you GTX260 works, then it will also work with the second application. It is just more work for us, that we need to maintain two queues. | |
ID: 13106 | Rating: 0 | rate: / Reply Quote | |
Yes, I would also prefer that my working gtx260 gets the "normal" work. You could just add an option to the preferences that says: "if your gtx260 doesn't work, check this!" | |
ID: 13110 | Rating: 0 | rate: / Reply Quote | |
Yes, I would also prefer that my working gtx260 gets the "normal" work. You could just add an option to the preferences that says: "if your gtx260 doesn't work, check this!" For me at least, such an option would need to be machine specific as I have 1 260 that runs without error and 1 that errors almost every WU. Of course, if there's no performance hit with the new 260 app, there'll be no need for the option. But, option or not, I'd be happy just to have both my 260s running again. and then there's my gtx295 with its GPU1 problems :( | |
ID: 13111 | Rating: 0 | rate: / Reply Quote | |
It would not matter if you GTX260 works, then it will also work with the second application. It is just more work for us, that we need to maintain two queues.But aren't we having 2 queues based on what you said? Clearly only some workunits can run there, but at least we move on without waiting. ____________ | |
ID: 13144 | Rating: 0 | rate: / Reply Quote | |
We are implementing a workaround, which is quite drastic. We will make an application just for 260 which does not use PME. Clearly only some workunits can run there, but at least we move on without waiting. Thanx for your work on this problem! I am waiting for this application for buggy 260s. ____________ From Siberia with love! | |
ID: 13167 | Rating: 0 | rate: / Reply Quote | |
I keep seeing references to this problem, but have never seen is on my own MSI 260. There's other people I know that are also running MSI without a problem. So does this bug only affect certain brands? Is it a problem with EVGA or are other brands affected? | |
ID: 13182 | Rating: 0 | rate: / Reply Quote | |
We are implementing a workaround, which is quite drastic. We will make an application just for 260 which does not use PME. Clearly only some workunits can run there, but at least we move on without waiting. Have you done a poll or run a query on the database to see how many 260s do not run correctly on the stock WU? Which mfg makes the 260s that do not run? If the erroring 260 are few, it doesn't make sense to me to make extraordinary effort to support them. Maybe those cards need a firmware upgrade. My MSI 260s run just fine. | |
ID: 13189 | Rating: 0 | rate: / Reply Quote | |
I have 2 EVGA GTX 260 216 that are 55nm and run flawless. | |
ID: 13193 | Rating: 0 | rate: / Reply Quote | |
This could happens, I have similar experience. Was trying several times to add 4th card to perfect working trinity of ASUS GTX 260's (216C, 55 nm). 4th card was reference Nvidia (192's , 65nm). | |
ID: 13200 | Rating: 0 | rate: / Reply Quote | |
I have 2 x GTX260 (Palit brand, 216 sp). They worked fine using 182.50 drivers. After upgrading drivers they don't. 185.xx, 186.xx and 190.xx drivers don't work with GPUgrid. | |
ID: 13204 | Rating: 0 | rate: / Reply Quote | |
I have a GTX260/216 (55nm) (MSI) that runs GPUGrid with no problems and a GTX260/216 (65nm)(EVGA) that will not run either GPUGrid or Collative without an error. Both run 190.38 drivers. | |
ID: 13205 | Rating: 0 | rate: / Reply Quote | |
6 BFG 260 216 and 2 EVGA 260 216 without any issues. | |
ID: 13223 | Rating: 0 | rate: / Reply Quote | |
| |
ID: 13225 | Rating: 0 | rate: / Reply Quote | |
Has anyone had success running a 192 core 260 on Linux? Mine were on Vista64, thinking about trying Linux. | |
ID: 13226 | Rating: 0 | rate: / Reply Quote | |
I have a 260 216 and it does not work under vistea 64 and win7 64. So I changed as many others to collatz. sorry. | |
ID: 13230 | Rating: 0 | rate: / Reply Quote | |
I have 2x BFG GTX260-216, both running linux | |
ID: 13232 | Rating: 0 | rate: / Reply Quote | |
I have a 260 216 and it does not work under vistea 64 and win7 64. So I changed as many others to collatz. What brand is the card, possibly EVGA? | |
ID: 13238 | Rating: 0 | rate: / Reply Quote | |
I have a ASUS GTX 260 216sp and it does not work under win7 64. So I changed as many others to collatz. | |
ID: 13240 | Rating: 0 | rate: / Reply Quote | |
I have got an ASUS GTX 260/216 Top (factory OC) 65nm, worked fine using 182.06 drivers. After upgrading to 185.xx, 186.xx, 190.xx and 191.xx drivers doesn't work with GPUgrid. Crunching Seti CUDA VLARkillv12, Collatz, MW and Folding works fine. | |
ID: 13245 | Rating: 0 | rate: / Reply Quote | |
I have a 260 216 and it does not work under vistea 64 and win7 64. So I changed as many others to collatz. That is a GAINWARD-card | |
ID: 13267 | Rating: 0 | rate: / Reply Quote | |
We are implementing a workaround, which is quite drastic. We will make an application just for 260 which does not use PME. Clearly only some workunits can run there, but at least we move on without waiting. This staement was from1oth October. Is the new application for the 260 available now? | |
ID: 13298 | Rating: 0 | rate: / Reply Quote | |
Our experience with gainward was very bad. The one we tried had a different dissipation system which were working very poorly with temperatures above of 100 degrees. These were GTX275. | |
ID: 13338 | Rating: 0 | rate: / Reply Quote | |
i WAS getting errors with my GTX260 until i STOPPED overclocking my CPU | |
ID: 13388 | Rating: 0 | rate: / Reply Quote | |
Our experience with gainward was very bad. The one we tried had a different dissipation system which were working very poorly with temperatures above of 100 degrees. These were GTX275. Ok, then still waiting. ____________ "Silakka" Hello from Turku > Åbo. | |
ID: 13392 | Rating: 0 | rate: / Reply Quote | |
Waiting as yet! | |
ID: 13515 | Rating: 0 | rate: / Reply Quote | |
Seems that volunteer's questions are remaining without response here... It's OK, we are still waiting... | |
ID: 13599 | Rating: 0 | rate: / Reply Quote | |
Still waiting too........ | |
ID: 13634 | Rating: 0 | rate: / Reply Quote | |
I sold my problem GTX 260 192 core 65nm. Kept the 2 55nm that do fine. The older 65nm cards are a problem for all but games. When new cards come out, like Fermi, and are available the prices will drop so the GTX 260 will be worth less. So sell while it is still worth a decent price. | |
ID: 13635 | Rating: 0 | rate: / Reply Quote | |
There are many GTX260 cards. Not all are good and not all are bad. | |
ID: 13650 | Rating: 0 | rate: / Reply Quote | |
Well, I have tried to run a couple WUS without overclocking, without using PC at all and with speedup videocard's fan from 90 to 100% (note, that usually it rotates at 40% of maximum speed), so as with maximum air flow through my PC's case (3R Systems Air 400, Hiper PS). But there was always the same famous fft-error. | |
ID: 13658 | Rating: 0 | rate: / Reply Quote | |
Just a note about fast fans on your card. I have an ATI 4850 crunching folding@home. It was running quite hot and the fan started to go round very fast. Then there were errors reported saying that the fan was causing vibrations and the project stopped running. | |
ID: 13678 | Rating: 0 | rate: / Reply Quote | |
Many cards take little account of any heat flow inside a typical case. | |
ID: 13679 | Rating: 0 | rate: / Reply Quote | |
There are still errors... | |
ID: 13694 | Rating: 0 | rate: / Reply Quote | |
use GPU-Z to see what the temperatures are doing. | |
ID: 13696 | Rating: 0 | rate: / Reply Quote | |
My Palit GTX260 216sp has successfully completed the last 33 GPUGRID tasks in a row - about 2 weeks work, and on a Vista Ult 64bit machine! | |
ID: 13749 | Rating: 0 | rate: / Reply Quote | |
Would it be reasonable to ask the BOINC developers to add code to report more details about what specific type of chip the GPU board uses, in order to help gather more information about which GTX260 boards work with GPUGRID? Or would it be more reasonable to add such code to your application instead, so that even most failed workunits would be able to send back such information? | |
ID: 14175 | Rating: 0 | rate: / Reply Quote | |
I think that in the long term it will be essential for most projects to be able to better determine hardware and allocate tasks accordingly. | |
ID: 14184 | Rating: 0 | rate: / Reply Quote | |
I still got 80 percent errors on my 260s. Maybe the developers should have a quiet talk with the guys from collatz? After wasting some time here i am again back to collatz. Will check again in 30 days. | |
ID: 15154 | Rating: 0 | rate: / Reply Quote | |
Change your project application preferences, you are using the old application, as I was! | |
ID: 15163 | Rating: 0 | rate: / Reply Quote | |
The problem with CUDA FFT on some 260 cards is still there. The new application does not solve it. We are working on eliminating the CUDAFFT all together. | |
ID: 15165 | Rating: 0 | rate: / Reply Quote | |
If the ACEMD 2 project runs 60% faster then there would be less chance of any error for any given task. Therefore any given task is more likely to complete and overall more work will be done by users that experience FFT errors. | |
ID: 15168 | Rating: 0 | rate: / Reply Quote | |
It would not matter if you GTX260 works, then it will also work with the second application. It is just more work for us, that we need to maintain two queues. An alternate plan to consider: put both application programs in each workunit, along with a script to choose which one of them to actually use. If the 260-only program can still handle all workunits, that should eliminate the second queue. A questionable idea if the 260-only application gives enough difference in outputs, though. | |
ID: 15420 | Rating: 0 | rate: / Reply Quote | |
Does new 6.15 beta application for Linux solve the bug? | |
ID: 15585 | Rating: 0 | rate: / Reply Quote | |
Any progress on the GTX 260 issue? Resolved? Still working on it? Not going to be fixed? | |
ID: 16530 | Rating: 0 | rate: / Reply Quote | |
As far as I know there has been no progress on this front at GPUGrid, however there is another application in the final stages of development, 6.72; it is possible that this may resolve some of the issues crunching with the older 65nm versions of the GTX 260 (effects the older GT200 cards, but not the newer GT200b). | |
ID: 16531 | Rating: 0 | rate: / Reply Quote | |
Nvidia told us that the problem is solved with CUDA3, but we did not check it. | |
ID: 16536 | Rating: 0 | rate: / Reply Quote | |
Nvidia told us that the problem is solved with CUDA3, but we did not check it. I have checked it, 6.72 tasks erroed out even on the underclocked one. Time to time a short test task finished OK. Host ID 31329. | |
ID: 16550 | Rating: 0 | rate: / Reply Quote | |
That's delivering cuda3 only to Fermi cards. | |
ID: 16551 | Rating: 0 | rate: / Reply Quote | |
Sorry, short test 3.0 tasks you have sent to GTX260 a weeks ago are deleted. I haven´t noticed it. Thanks for explaining of using 3.0 for Fermi cards only. | |
ID: 16568 | Rating: 0 | rate: / Reply Quote | |
so is this problem fixed yet? i have two gtx260's, one is working fine, the other one is a 65nm core216 and doesn't work. please let me know if there's hope in holding on to this card for a future fix, or if i should just sell this card because there won't be any plans to problem. | |
ID: 17009 | Rating: 0 | rate: / Reply Quote | |
Sorry, short test 3.0 tasks you have sent to GTX260 a weeks ago are deleted. I haven´t noticed it. Thanks for explaining of using 3.0 for Fermi cards only. The reason is that Fermi require 3.0 tasks, while for other cards the code compiled with 3.0 is slower. I can change the server to deliver also to GTX260. gdf | |
ID: 17022 | Rating: 0 | rate: / Reply Quote | |
| |
ID: 17024 | Rating: 0 | rate: / Reply Quote | |
This will be interesting! | |
ID: 17028 | Rating: 0 | rate: / Reply Quote | |
The reason is that Fermi require 3.0 tasks, while for other cards the code compiled with 3.0 is slower. Hopefully that won't send the slower code to all the good GTX 260 cards too. Slowing them all down to fix just a few is not so good. You could possibly make a new choice in preferences with a different queue: ACEMD: no ACEMD ver 2.0: no ACEMD beta: no GTX260/65nm only: yes or ACEMD: no ACEMD ver 2.0: no ACEMD beta: no Fermi / GTX260/65nm only: yes | |
ID: 17031 | Rating: 0 | rate: / Reply Quote | |
so are these updated to accept fermi WU's yet? | |
ID: 17224 | Rating: 0 | rate: / Reply Quote | |
Fermis already work. | |
ID: 17228 | Rating: 0 | rate: / Reply Quote | |
I am Ubuntu 32 bit with 260-216--does it not run on 32 bit?? | |
ID: 17237 | Rating: 0 | rate: / Reply Quote | |
gpugrid is 64 bit only on Linux. Time to upgrade anyways. 64 bit is where its at. | |
ID: 17346 | Rating: 0 | rate: / Reply Quote | |
The reason is that Fermi require 3.0 tasks, while for other cards the code compiled with 3.0 is slower.I recieve cuda (not cuda30) tasks for my unusable 65nm GTX260. Could you add a checkbox to choose standard cuda23 or cuda30 for non-Fermi class GPU? | |
ID: 17541 | Rating: 0 | rate: / Reply Quote | |
I would like to buy the new gtx460 to use on gpugrid.. will it work? | |
ID: 17969 | Rating: 0 | rate: / Reply Quote | |
i have a gtx260 that always failed the tasks.. | |
ID: 18138 | Rating: 0 | rate: / Reply Quote | |
Your GTX260 can only run 6.05 tasks, so it does not use CUDA3.1. You also have the 192 shader version, which tends to be much more error prone. | |
ID: 18139 | Rating: 0 | rate: / Reply Quote | |
You are using Windows 7 Ultimate x64. I found that almost all tasks failed on my GTX260 when I used it with Windows 7 x64, so I installed Win XP x86 and now almost all tasks complete successfully. I also ran FatPuppy Linux on the card for abou 10days, and again virtually all tasks completed. The above advice is irrelevant. SK, he's using the 192 shader GTX 260 with the bug and yours is the 216 shader version. Why you ever had problems with your 216 shader GTX 260 is mysterious. It should run fine with any OS. | |
ID: 18140 | Rating: 0 | rate: / Reply Quote | |
Is the project working along to solve this?? | |
ID: 18620 | Rating: 0 | rate: / Reply Quote | |
6.11 tasks (CUDA 3.1) errored out on GTX 260 65nm - rev. A. Win XP x_64, driver 2650.99 | |
ID: 19172 | Rating: 0 | rate: / Reply Quote | |
Is the project working along to solve this?? It's not a problem the project can fix by itself. It has been an issue from the outset with some of the 65nm GTX260's cards and could only be fixed by a driver and through the CUDA developers app, both written by NVidia. My guess is that the odd 65nm GTX260-192 that works may have newer firmware/ use a different revision. lkiller123, what revision is your card and have you tried to run tasks at reference speeds? I think there is a snowball's chance in hell of a 65nm OC'd GTX260-192 crunching GPUGrid tasks in the same system as another OC'd GTX260. In the past I struggled with 2 such cards and could get neither to work here. One ran MW tasks reasonably well, failing only the odd one, and I think worked on Folding but the other card was a deal loss. | |
ID: 19173 | Rating: 0 | rate: / Reply Quote | |
Is there a way to update the firmware? My limping GTX260 is unusable here and I'm willing to try everything to get it working here. | |
ID: 19174 | Rating: 0 | rate: / Reply Quote | |
Basically No. As far as I am aware the Bios (firmware/instruction set) is fixed at the factory and NVidia want to keep it that way. | |
ID: 19175 | Rating: 0 | rate: / Reply Quote | |
Here are the GPU-Z shots of both cards: | |
ID: 19183 | Rating: 0 | rate: / Reply Quote | |
Found a possible utility to update some cards. | |
ID: 19193 | Rating: 0 | rate: / Reply Quote | |
Thanks for the video. I will give it a shot sometime soon. | |
ID: 19203 | Rating: 0 | rate: / Reply Quote | |
Message boards : Graphics cards (GPUs) : Is the problem whit GTX 260, Ok now??