Author |
Message |
|
Incorrect function. (0x1) - exit code 1 (0x1)
WU #1
WU #2
Ok here is my setup
Q6600 @ 3.0 Ghz
gtx 260 core 216 superclocked Driver: 180.48
8600GTS 512Mb (no OC)Driver: 178.13
Win XP SP3
Work units so far have been successful on the GTX 260 but on the 8600GTS they are erroring out...
Any Hints/Suggestions? |
|
|
Kokomiko Send message
Joined: 18 Jul 08 Posts: 190 Credit: 24,093,690 RAC: 0 Level
Scientific publications
|
Any Hints/Suggestions?
Why do you use different drivers on one PC for 2 cards? Both should run wih the same driver and I would prefer the 178.24.
____________
|
|
|
|
Yeah, I agree with Kokomiko. I am starting to wonder if GPUGrid is starting to issue WU's based on card configuration. BOINCmgr knows that you have a 260 and an 8600, downloads two different WU's to assign to each one of your cards. but when downloaded, just send it to the first available card which may not be the correct recipient for the WU.
They are sending WU's of varying size to allow weaker cards the ability to finish a WU in under 4 days.
Any other ideas?
Pat |
|
|
|
I updated the drivers to the newest available for the card.. no specific reason actually...
Boinc seems to think I have 2 8600GTS's in the start-up messages
12/19/2008 5:28:23 PM||CUDA devices found
12/19/2008 5:28:23 PM||Coprocessor: GeForce 8600 GTS (2)
I will wait and see what the current 2 running will come up with...
~Bob |
|
|
|
Hense BOINCmgr has now choosen the weaker of the two cards so that both cards can process the smaller WU's. Still a theory though.
Pat |
|
|
|
Hense BOINCmgr has now choosen the weaker of the two cards so that both cards can process the smaller WU's. Still a theory though.
I don't think it's that smart yet.
MrS
____________
Scanning for our furry friends since Jan 2002 |
|
|
|
Well the next 3 tasks finished.
The 2 that ran on the gtx 260 again finished fine,
The one on the 8600gts didn't...
Here is what the log says...
12/20/2008 3:16:42 AM|GPUGRID|Computation for task Ze27319-SH2_US_1-6-40-SH2_US_12150000_1 finished
12/20/2008 3:16:43 AM|GPUGRID|Starting xN19912-SH2_US_1-6-40-SH2_US_1120000_1
12/20/2008 3:16:44 AM|GPUGRID|Starting task xN19912-SH2_US_1-6-40-SH2_US_1120000_1 using acemd version 655
12/20/2008 3:16:45 AM|GPUGRID|Started upload of Ze27319-SH2_US_1-6-40-SH2_US_12150000_1_0
12/20/2008 3:16:45 AM|GPUGRID|Started upload of Ze27319-SH2_US_1-6-40-SH2_US_12150000_1_1
12/20/2008 3:17:02 AM|GPUGRID|Finished upload of Ze27319-SH2_US_1-6-40-SH2_US_12150000_1_0
12/20/2008 3:17:02 AM|GPUGRID|Started upload of Ze27319-SH2_US_1-6-40-SH2_US_12150000_1_2
12/20/2008 3:17:21 AM|GPUGRID|Finished upload of Ze27319-SH2_US_1-6-40-SH2_US_12150000_1_1
12/20/2008 3:17:21 AM|GPUGRID|Started upload of Ze27319-SH2_US_1-6-40-SH2_US_12150000_1_3
12/20/2008 3:17:25 AM|GPUGRID|Finished upload of Ze27319-SH2_US_1-6-40-SH2_US_12150000_1_3
12/20/2008 3:17:33 AM|GPUGRID|Finished upload of Ze27319-SH2_US_1-6-40-SH2_US_12150000_1_2
12/20/2008 3:20:47 AM|GPUGRID|Computation for task Yd21241-SH2_USPME-2-40-SH2_USPME1910000_0 finished
12/20/2008 3:20:47 AM|GPUGRID|Output file Yd21241-SH2_USPME-2-40-SH2_USPME1910000_0_1 for task Yd21241-SH2_USPME-2-40-SH2_USPME1910000_0 absent
12/20/2008 3:20:47 AM|GPUGRID|Output file Yd21241-SH2_USPME-2-40-SH2_USPME1910000_0_2 for task Yd21241-SH2_USPME-2-40-SH2_USPME1910000_0 absent
12/20/2008 3:20:47 AM|GPUGRID|Output file Yd21241-SH2_USPME-2-40-SH2_USPME1910000_0_3 for task Yd21241-SH2_USPME-2-40-SH2_USPME1910000_0 absent
12/20/2008 3:20:49 AM|GPUGRID|Started upload of Yd21241-SH2_USPME-2-40-SH2_USPME1910000_0_0
12/20/2008 3:20:54 AM|GPUGRID|Finished upload of Yd21241-SH2_USPME-2-40-SH2_USPME1910000_0_0
task Ze27319-SH2_US_1-6-40-SH2_US_12150000_1 was on gtx 260
the other on the 8600GTS
So basically what I gather from the logs is that it finished fine yet the output was missing, yet it uploaded it anyway? |
|
|
|
No other suggestions? |
|
|
|
OK, let's sort this information a bit and see what we find:
- all tasks which run only on the GTX 260 are fine
- all tasks which run only on the 8600GTS fail
- one task started by GTS and finished by GTX is also fine
- the tasks fail after a certain time, not immediately
-> i.e. your setup works in principle
The message "incorrect function" means that something went wrong during calculation, which can be caused e.g. by overclocking or bad cards. The task fails an therefore no output file is written. Something is uploaded, but I think this is basically the error message. The missing output file is not causing the error, it's a symptom.
Could it be that your 8600GTS overheats? The coolers on such "low end" cards are often cheap and just barely strong enough. Plus there may be some dust already, heat from the GTX and the card / chip could already be a bit older (and be thus less tolerant).
MrS
____________
Scanning for our furry friends since Jan 2002 |
|
|
|
The temps on the 8600gts range from 75-90 C under full load... (depending on what's running)
A little warm yes.. But still below the 105 C limit...
No over-clock on the card... (wouldn't even think about it with the temps it has already)
Possibly a part of the card overheats while the rest doesn't?
Would you recommend a re-application of the cooling/thermal paste? |
|
|
|
If it's possible I would test the PC with the side panel open. Watch if the temps improve when another WU runs. If that one succeds we can be pretty sure we found the issue. Additionally you could log the temps, e.g. with RivaTuner. There should be a sharp drop in temperature when a WU fails and we can take a look at the temperature just before failure.. maybe it reached 95°C at this point? (just an example)
I wouldn't bother reapplying the thermal grease. That should give you 1 or 2 degrees, nothing game-changing. Dust in the cooler / fan is much more dangerous.
And I'd rather mount a nice 3rd party cooler, if I had taken off the old one anyway (to reapply the paste).
MrS
____________
Scanning for our furry friends since Jan 2002 |
|
|
|
Those temperatures were with the case open :P (both sides off)My case is really bad for airflow...
Unfortunately (Fortunately?) my drivers got all corrupt (thanks to seti's cuda) and during the troubleshooting to find out what the problem was I removed the 8600GTS... Most likely will re-install during the new year... |
|
|
|
90°C with the case open? That must be one hell of a bad cooler.. or it's set to spin low and stay relatively silent. Do you know how to set fan speeds yourself?
MrS
____________
Scanning for our furry friends since Jan 2002 |
|
|
|
The fans were left at the stock auto settings...
I have yet to find a program that can modify the second gpu...Do you know of any?
Unless of course there is a way to set it that I'm not aware of? |
|
|
K1atOdessaSend message
Joined: 25 Feb 08 Posts: 249 Credit: 387,028,788 RAC: 1,197,795 Level
Scientific publications
|
I have yet to find a program that can modify the second gpu...Do you know of any?
I use Rivatuner 2.22 and it will control the fan speeds of multiple gpu's. |
|
|
|
I use Rivatuner on my box with multiple GPUs - I think installing nTune will also give you the same capability - although Riva is a much more comprehensive tool. EVGA Precision is fine for boxes with a single GPU.
P. |
|
|