Hello all.

When I first time bought my new GTX460 (Gigabyte 1Gb, factory OC (715MHz Core/1800MHz Memory)) it was working nice and properly with cuda fermi tasks.
But after some time I found that ALL my cuda fermi tasks are finishing with "Calculation Error" shown in BOINC manager Tasks/Status field. The calculation time is around 1-2 seconds for each task. I could remeber only Graphics driver change to the latest from Nvidia (WHQL) and perhaps BOINC manager change to 6.10.58
What can be the reason and how to solve this issue?
Thank you

Those I checked have finished with "CreateProcess() failed - Access is denied. (0x5)".

Gruß,
Gundolf

---

Computer sind nicht alles im Leben. (Kleiner Scherz)

SETI@home classic workunits 3,758
SETI@home classic CPU time 66,520 hours

Thank you - I guess so. But what has happened? And what to do?

Try to re-install BOINC
Select "Permit all users to control BOINC"
(I hope this will set correct permissions on the directories used by BOINC)

Did you reboot recently?

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

Ok, thank you. I will reinstall the latest version with "permit all users".
I am rebooting quite periodically, and definetely made it several times when this problem appeared. Unfortunately there is no CUDA work now, so I should wait after reinstall.

Interesting - today evening I came home and found out that rosetta project started to make "computation errors".
Now I uninstalled and than installed BOINC once again with "permit all users". I'll see the difference.

10/11/2010 00:32:29 rosetta@home [error] Process creation failed: Access is denied. (0x5)

This message appears immediately after total delete and clean install - with "allow all users to control BOINC"

For SETI - could not check now as it is "down for maintenance".
I will wait for a while.

One more thing - I found out that there are local groups boinc_admins and boinc_projects, but there NO users boinc_admin and boinc_project.
I created them manually and will look how it will work.

It is strange because on my second computer they exist.

Yes, I think I found the reason. It was antivirus program which put Rosetta executable into the "untrusted applications" without asking for decision.
I changed the properties to "trusted" and now Rosetta is working.

I hope SETI project has the same issue. I will check it after new jobs will be available.

Yes, I think I found the reason. It was antivirus program which put Rosetta executable into the "untrusted applications" without asking for decision.
I changed the properties to "trusted" and now Rosetta is working.

I hope SETI project has the same issue. I will check it after new jobs will be available.

I was about to write "switch off your antivirus program for a few minutes to see if that helps"
and you found it yourself :)

Now, to help others when they have the same problem
will you tell us which antivirus program makes such things?

( I never have such problems with NOD32 Antivirus (several versions 2.70 - 4.x on several computers/OSes) (used also AVG, Avast in the past)
Nothing is in ignore (exclude) list of NOD32.
)

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

Yes, this is great.

There are still no SETI jobs available, but I think this will help.

It was Kaspersky Internet Security 2011 (Trial) and it has self firewall with some kind of "stealth" decisions for some programs (to allow them or not to run or create files). Perhaps it can be tuned by basically it brought me to a mess successfully fixed.

By the way, my system:
Win7 Ultimate 32bit, AMD 1090T, ASUS M4N75TD, Gigabyte GTX460 1Gb, 4 GB RAM (3 GB usable), Western Digital WD1001FALS 1Tb

Hello once again.

There are still no SETI jobs available, and I tried Milky way project with CUDA jobs ("cuda23"). Strange name as lately it was "cuda_fermi" remark.
Unfortunately the result was the same - 100% progress with "Calculation error" remark in "Jobs" section. But no suspicious messages - only:
"14.11.2010 12:37:54 Milkyway@home Starting de_separation_82_2s_40_2_1837301_1289731047_0

14.11.2010 12:37:54 Milkyway@home Starting task de_separation_82_2s_40_2_1837301_1289731047_0 using milkyway version 24

14.11.2010 12:37:55 Milkyway@home Computation for task de_separation_82_2s_40_2_1837301_1289731047_0 finished"



1. I tuned at first and then completely disable antivirus software - it did not help to solve the problem.

2. Now I am trying to downgrade the nVidia Video driver from 260.99 WHQL (the latest) to 258.96 WHQL.

No - the same result.

3. I thought it was power save functions of my GTX460 - in standby (windows desktop) mode the frequencies fell down to 51MHz core and 135MHz memory.
No - the same result.

4. I will continue to downgrade the driver version to find the reason.

The attempt to use 258.56 driver coming with the bundled CD make no positive effect: the result was the same - 100% progress with "Calculation error" remark in "Jobs" section. But no suspicious messages - only:

"14.11.2010 18:00:59 Milkyway@home Starting de_separation_82_3s_00_2_8955_1289750263_0
14.11.2010 18:00:59 Milkyway@home Starting task de_separation_82_3s_00_2_8955_1289750263_0 using milkyway version 24
14.11.2010 18:01:00 Milkyway@home Computation for task de_separation_82_3s_00_2_8955_1289750263_0 finished"

Now I have no ideas of the reasons of "Calculation error" with GPU jobs.
Do anybody has any?

Since your Milkyway account doesn't show here yet, I can't check, but if you locate a task with "Calculation error" remark in "Jobs" section, click on the task ID link and look for error messages there.

Gruß,
Gundolf

I caught the "Computation error" from GPU job from MilkyWay project.
I can find the Properties only in the "Task" tab of BOINC manager.
It has several rows. How could I paste a picture here? Can I attach it, because Paste is not working.

No, not in the manager, online.

Go to your account (at Milkyway) and select "Tasks View". On the following page, select the ID of an erroneous task.

Gruß,
Gundolf

Thanks.

Here it is:

Name de_separation_82_2s_40_2_335086_1289835580_0
Workunit 186255761
Created 15 Nov 2010 15:39:50 UTC
Sent 15 Nov 2010 15:44:15 UTC
Received 15 Nov 2010 15:49:49 UTC
Server state Over
Outcome Client error
Client state Compute error
Exit status 1 (0x1)
Computer ID 234269
Report deadline 23 Nov 2010 15:44:15 UTC
Run time 0
CPU time 0
stderr out

<core_client_version>6.10.58</core_client_version>
<![CDATA[
<message>
Incorrect function. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
Device index specified on the command line was 0
Looking for a Double Precision capable NVIDIA GPU
The device GeForce GTX 460 from the command line cannot be used because a device supporting compute capability 1.3 (Double Precision) is required
Found 1 CUDA cards
Found a GeForce GTX 460
Device cannot be used, it does not have compute capability 1.3 support
No compute capability 1.3 cards have been found, exiting...

</stderr_txt>
]]>

Validate state Invalid
Claimed credit 0
Granted credit 0
application version MilkyWay@Home v0.24 (cuda23)



Can it help? Why my card has "no capability"?

My full video card name is
GV-N460OC-1GI

http://gigabyte.com/products/product-page.aspx?pid=3530

Built-in video (ASUS motherboard, M4N75TD, http://www.asus.com/product.aspx?P_ID=b6R5QXWNOAJzwcUx&templete=2) is disabled, so only 1 CUDA processor is active. There are no problems in games. Just finished "Medal of Honor 2010" on the maximum quality - no sign of ANY picture damages, hangs, slow downs or any other issues.

I am worrying if I had really seen my card working under CUDA jobs correctly :-)

How could I paste a picture here? Can I attach it, because Paste is not working.

I use this site to upload:
http://ploader.net/

(no registration needed; use .png for image format)

Copy (to your post) the last (#3) link that ploader.net will generate (after upload),
mark the line (link) ([Home] [Shift-End])
and press {Img} button above (near the {URL} button)

e.g.:
[img]http://ploader.net/files/963a20a09afb150f29996b4ccd820e52.png[/img]

will give:




(to see how someone uses tags to achieve something press {Quote} button under the post -
e.g. see how I disabled the recognition of [img]...[/img] as tags and instead it is shown as text)

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

Thanks.

Here it is:

Name de_separation_82_2s_40_2_335086_1289835580_0
Workunit 186255761
Created 15 Nov 2010 15:39:50 UTC
Sent 15 Nov 2010 15:44:15 UTC
Received 15 Nov 2010 15:49:49 UTC
Server state Over
Outcome Client error
Client state Compute error
Exit status 1 (0x1)
Computer ID 234269
Report deadline 23 Nov 2010 15:44:15 UTC
Run time 0
CPU time 0
stderr out

<core_client_version>6.10.58</core_client_version>
<![CDATA[
<message>
Incorrect function. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
Device index specified on the command line was 0
Looking for a Double Precision capable NVIDIA GPU
The device GeForce GTX 460 from the command line cannot be used because a device supporting compute capability 1.3 (Double Precision) is required
Found 1 CUDA cards
Found a GeForce GTX 460
Device cannot be used, it does not have compute capability 1.3 support
No compute capability 1.3 cards have been found, exiting...

</stderr_txt>
]]>

Validate state Invalid
Claimed credit 0
Granted credit 0
application version MilkyWay@Home v0.24 (cuda23)


Can it help? Why my card has "no capability"?

According to this table GeForce GTX 460 have to have even Compute Capability 2.0
http://www.geeks3d.com/20100606/gpu-computing-nvidia-cuda-compute-capability-comparative-table/

Maybe "application version MilkyWay@Home v0.24" can't run on Fermi GPUs??

The task did not run at all - it immediately crashed (Run time 0) at 0% (not 100%)
but every crashed or aborted tasks go from 0 to 100% "in split second".

(I don't run MilkyWay@Home, read on their site what hardware they need/support)

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)