Author |
Message |
DestiSend message
Joined: 10 Jul 07 Posts: 19 Credit: 1,272,950 RAC: 0 Level
Scientific publications
|
Hello,
I tried to restart some GPUGRID work on my 9600GT, but all workuntis failed. Some of them immeditaly fail and some are running for hours and then become invalid.
I wonder if GPUGRID had dropped support for that chip?
http://www.gpugrid.net/results.php?userid=165
OpenGL vendor string: NVIDIA Corporation
OpenGL renderer string: GeForce 9600 GT/PCI/SSE2
OpenGL version string: 3.2.0 NVIDIA 195.36.24
____________
Linux Users Everywhere @ BOINC |
|
|
|
Have you got the latest drivers? |
|
|
skgivenVolunteer moderator Volunteer tester
Send message
Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level
Scientific publications
|
A GT 9600 is on the slow side; with only 64 shaders it is going to take about a day or so to complete one task running 24/7.
Some of your tasks seem to fail after a few seconds, but this is the case with many other cards, and does not effect overall performance too much.
Two of your tasks failed after a longer time, about 8hours. This suggests your card is capable of running GPUGRID tasks; it may have failed all tasks, but if it was totally incapable it is unlikely to have failed two tasks after 8hours runtime; more likely much sooner.
We don’t know much else about what you do on the system - crunch other tasks on the CPU or GPU, watch films, online streaming content or play games. These things can all lead to failures.
You might want to try a newer driver, 195.36.24 is quite old.
I see that you have 2 tasks in progress. Given that these tasks are likely to take a day on your card it would be wise to reduce the cache to 0.1days.
|
|
|
DestiSend message
Joined: 10 Jul 07 Posts: 19 Credit: 1,272,950 RAC: 0 Level
Scientific publications
|
It seems there were a little too less free space, I moved it to another partiton and now it's working better, but sometimes I still got the memalloc fail when I had stopped the workunit and restart it. I will check if it's fixed with the 256. drivers.
____________
Linux Users Everywhere @ BOINC |
|
|
jlhalSend message
Joined: 1 Mar 10 Posts: 147 Credit: 1,077,535,540 RAC: 0 Level
Scientific publications
|
Hi !
If you click on my nick in gpugrid you wiil see I' m also using 9600GT :
2 GPUs : 9600GT TurboForce (Latest 256.xx drivers from Nvidia)
1 CPU : AMD 9350E Phenom X4
Windows XP Pro X64 Edition and 8GB memory.
I'm also crunching jobs for Grid Community (4 tasks concurrently) : every thing is fine, jobs never crash.
I must say I got sometimes a bunch of failed gpugrid jobs too, even if I do not accept test apps, everytime signaling problem to Microsoft !
But it may also run for a week without an error.
I' m still wondering why these jobs are crashing. I think there is something very badly handled in them !
Greetings from Paris... |
|
|
DestiSend message
Joined: 10 Jul 07 Posts: 19 Credit: 1,272,950 RAC: 0 Level
Scientific publications
|
Next workunit reaches 100% and then failed.
07-Aug-2010 16:51:40 [GPUGRID] Computation for task I238-TONI_KIDln-50-100-RND8527_0 finished
07-Aug-2010 16:51:40 [GPUGRID] Output file I238-TONI_KIDln-50-100-RND8527_0_1 for task I238-TONI_KIDln-50-100-RND8527_0 absent
07-Aug-2010 16:51:40 [GPUGRID] Output file I238-TONI_KIDln-50-100-RND8527_0_2 for task I238-TONI_KIDln-50-100-RND8527_0 absent
07-Aug-2010 16:51:40 [GPUGRID] Output file I238-TONI_KIDln-50-100-RND8527_0_3 for task I238-TONI_KIDln-50-100-RND8527_0 absent
http://www.gpugrid.net/result.php?resultid=2779446
____________
Linux Users Everywhere @ BOINC |
|
|
|
Guys, can you take your shader and GPU core clocks back a bit and see if it changes anything? At least 27 MHz on the core and at least 54 MHz for the shaders.
MrS
____________
Scanning for our furry friends since Jan 2002 |
|
|
jlhalSend message
Joined: 1 Mar 10 Posts: 147 Credit: 1,077,535,540 RAC: 0 Level
Scientific publications
|
A GT 9600 is on the slow side; with only 64 shaders it is going to take about a day or so to complete one task running 24/7 ...
Not so much indeed for a NX96T1GHP : varying between 8 and 15 hours, giving from around 5000 ro 15000 credit !!! ..
Cheers |
|
|
|
It takes you 78300s to finish a 5700 credit WU (+bonus). That's just shy of 1 day. A 13300 credit WU took you 160700s, that is 1day 20h 38min. There's nothing wrong with these values or your card - it's just what the 9600GT can do.
By the way: you do get quite a few errors. Ar first glance they seem to be random and appear after quite some crunching time. This could be some systematic incompatibility between the old chips and the current software (drivers, app) or it could be that your chip has aged so far that it can't take the 1.80 GHz any more, or at least sometimes fails.
MrS
____________
Scanning for our furry friends since Jan 2002 |
|
|
jlhalSend message
Joined: 1 Mar 10 Posts: 147 Credit: 1,077,535,540 RAC: 0 Level
Scientific publications
|
...
By the way: you do get quite a few errors. Ar first glance they seem to be random and appear after quite some crunching time. This could be some systematic incompatibility between the old chips and the current software (drivers, app) or it could be that your chip has aged so far that it can't take the 1.80 GHz any more, or at least sometimes fails.
MrS
Hi ! Think you're right but does someone can explain why the folowing error occurs:
<MDIO ERROR: cannot open file "restart.coor">
When (in terms of context), does gpugrid need to open this file ?
Greetings |
|
|
|
Don't worry about that, it is not really an error. You will see that same line on all WUs even when they complete sucessfully.
____________
Thanks - Steve |
|
|
jlhalSend message
Joined: 1 Mar 10 Posts: 147 Credit: 1,077,535,540 RAC: 0 Level
Scientific publications
|
Don't worry about that, it is not really an error. You will see that same line on all WUs even when they complete sucessfully.
Thanks Snow Crash !
By the way, after rebooting Win XP, I could get the exact Boinc messages about my GPUs:
NVIDIA GPU 0: GeForce 9600 GT (driver version 25896, CUDA version 3010, compute capability 1.1, 1024MB, 230 GFLOPS peak)
NVIDIA GPU 1: GeForce 9600 GT (driver version 25896, CUDA version 3010, compute capability 1.1, 1024MB, 230 GFLOPS peak)
Greetings...
PS: a Gigabyte GTX460 1GB is planned for the middle of this week...
|
|
|
jlhalSend message
Joined: 1 Mar 10 Posts: 147 Credit: 1,077,535,540 RAC: 0 Level
Scientific publications
|
.
...
PS: a Gigabyte GTX460 1GB is planned for the middle of this week...
Hi !
Just installed the GTX 460 replacing both 9600GT, here is the messge from BOINC for this GPU :
NVIDIA GPU 0: GeForce GTX 460 (driver version 25896, CUDA version 3010, compute capability 2.1, 1024MB, 641 GFLOPS peak)
Cheers
|
|
|
uBronan Send message
Joined: 1 Feb 09 Posts: 139 Credit: 575,023 RAC: 0 Level
Scientific publications
|
Since my 9600 gt only ends in error after 25 hours of work i stopped this project, and since i am not gonna ever buy a nvidia i think its the end here.
So good luck to you all. |
|
|
|
.
...
PS: a Gigabyte GTX460 1GB is planned for the middle of this week...
Hi !
Just installed the GTX 460 replacing both 9600GT, here is the messge from BOINC for this GPU :
NVIDIA GPU 0: GeForce GTX 460 (driver version 25896, CUDA version 3010, compute capability 2.1, 1024MB, 641 GFLOPS peak)
Cheers
19.09.2010 16:04:04 NVIDIA GPU 0: GeForce GTX 460 (driver version 26063, CUDA version 3020, compute capability 2.1, 993MB, 363 GFLOPS peak)
thats mine Gigabyta GTX460 (the factory OC one..) i still dont understand how BOINC calc the GFLOPS peak..
how can see/calc the real speed?
____________
|
|
|