For three days, all units go to error (paola), where does the problem? I specify that the units (IBUCH) calculated without problems!
Sorry for my English, I write it with google translation. thank you
My pc: Athlon II X3, GTX 550ti, ubuntu.

Ah im not the only one with this problem ^^ is it only with app 42 like me?

---

DSKAG Austria Research Team: http://www.research.dskag.at

Hi, yes, cuda 42 PAOLA. My graphics card is compatible with CUDA 4.2, I do not understand?

I've checked the error message of these 2HDQ_43_9-PAOLA_2HDQ workunits in their stderr outpuf file, and I've find out that every one of them on every host I've checked resulted in ERROR: Failed to parse input file. So I've came to the conclusion that the source of this error is not on your side. This batch of 2HDQ_43_9-PAOLA_2HDQ workunits are messed up somehow.

Oh okay, thank you for your answer!

it looks like it arranges, I have a unit (PAOLA) that runs without error for 27 minutes.

it looks like it arranges, I have a unit (PAOLA) that runs without error for 27 minutes.

This workunit comes from a different batch: 3EKO_8_10-PAOLA_3EKO_8LIGs

There's definitely problems with the PAOLA_2HDQ batch. These tasks all fail on all cards no matter which app is used (3.1 or 4.2). Fortunately they all fail after a few seconds:

Stderr output

<core_client_version>7.0.24</core_client_version>
<![CDATA[
<message>
process exited with code 98 (0x62, -158)
</message>
<stderr_txt>
ERROR: file mdsim.cpp line 167: Failed to parse input file
20:12:31 (5458): called boinc_finish

</stderr_txt>
]]>

http://www.gpugrid.net/workunit.php?wuid=3543365

name 2HDQ_32_2-PAOLA_2HDQ-0-100-RND5005
application ACEMD2: GPU molecular dynamics
created 4 Jul 2012 | 20:41:18 UTC
minimum quorum 1
initial replication 1
max # of error/total/success tasks 7, 10, 6
errors Too many errors (may have bug)

5571497 103419 4 Jul 2012 | 21:52:29 UTC 4 Jul 2012 | 21:58:43 UTC Error while computing 2.20 0.05 --- ACEMD2: GPU molecular dynamics v6.16 (cuda31)
5572053 111371 4 Jul 2012 | 23:30:37 UTC 5 Jul 2012 | 1:35:58 UTC Error while computing 2.20 0.12 --- ACEMD2: GPU molecular dynamics v6.16 (cuda42)
5572983 58172 5 Jul 2012 | 3:42:30 UTC 5 Jul 2012 | 11:00:52 UTC Error while computing 3.02 0.03 --- ACEMD2: GPU molecular dynamics v6.16 (cuda42)
5575434 95154 5 Jul 2012 | 13:01:46 UTC 5 Jul 2012 | 13:08:38 UTC Error while computing 2.55 0.12 --- ACEMD2: GPU molecular dynamics v6.16 (cuda31)
5575986 122075 5 Jul 2012 | 14:20:47 UTC 5 Jul 2012 | 19:13:43 UTC Error while computing 2.02 0.04 --- ACEMD2: GPU molecular dynamics v6.16 (cuda42)
5577441 128121 5 Jul 2012 | 23:15:57 UTC 5 Jul 2012 | 23:22:37 UTC Error while computing 2.30 0.11 --- ACEMD2: GPU molecular dynamics v6.16 (cuda31)
5578198 113979 6 Jul 2012 | 1:41:12 UTC 6 Jul 2012 | 1:48:00 UTC Error while computing 0.00 0.00 --- ACEMD2: GPU molecular dynamics v6.16 (cuda42)
5578603 24523 6 Jul 2012 | 6:10:45 UTC 6 Jul 2012 | 11:21:31 UTC Error while computing 43.44 0.25 --- ACEMD2: GPU molecular dynamics v6.16 (cuda42)

Thanks,

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Hi dear volunteers,

There was such a stupid error in the input file :p, I fixed it and now I am going to submit again the system (500 WU) on acemd short (12500 credits for WU), the group is 2HDQbis.

thanks for your patience and your computing time :D

Paola

This is not a parse file error, but this WU has failed on 2 other rigs. One was with 295 so maybe their end, but the other was with a 470 301.42 driver.

3EKO_36_9_step_13_20_7-PAOLA_ADAPT-10-20-RND9221


<core_client_version>7.0.25</core_client_version>
<![CDATA[
<message>
The system cannot find the path specified. (0x3) - exit code 3 (0x3)
</message>
<stderr_txt>
MDIO: cannot open file "restart.coor"
SWAN : FATAL : Cuda driver error 999 in file 'swanlibnv2.cpp' in line 1574.
Assertion failed: a, file swanlibnv2.cpp, line 59


Cheers

There are still so much errors that wasting energy on resent failed workunits :(

---

DSKAG Austria Research Team: http://www.research.dskag.at

Gotta take the good with the bad. This is the only WU that has crashed for me.

Things happen.

Cheers.

Agreed to take the bad with the good. Just had a 1E2I fail after 7 hours. I get occassional driver crashes and permanently frozen displays with the 4.2's. All 4.2's seem 'touch and go' regardless of the author. Any further refinements would be much appreciated.

Try this post or this post, among others, in the "FAQ - Why does my run fail? Some answers."

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

I have error with new PAOLA units "1H46 RNP" ?

I had 1H46_19_9-PAOLA_1H46_RNP-6-100-RND5172_1 fail with

<core_client_version>6.10.60</core_client_version>
<![CDATA[
<message>
The system cannot find the path specified. (0x3) - exit code 3 (0x3)
</message>
<stderr_txt>
MDIO: cannot open file "restart.coor"
SWAN : FATAL : Cuda driver error 999 in file 'swanlibnv2.cpp' in line 1574.
Assertion failed: a, file swanlibnv2.cpp, line 59

This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.

</stderr_txt>
]]>

Ran for 19.29 minutes

Hello,
Congratulations to all the team for your job.

All actual units (PAOLA) are exiting with error code 247, see 5747509 for instance.

Stderr output

<core_client_version>7.0.29</core_client_version>
<![CDATA[
<message>
process exited with code 247 (0xf7, -9)
</message>
<stderr_txt>

</stderr_txt>
]]>

All this units are using cuda42. The only cuda31 unit I got was going fine until I've done a mistake, using the gpu at the same time.

My card is a GTX260, on a freshly installed fedora 17 system, with NVIDIA driver 304.32 and cuda Toolkit 4.2.9. I have followed the procedure at http://doc.fedora-fr.org/wiki/Cuda. The GTK toolkit is working fine and primegrid also.

What are this errors due to ?
Thank you for your help

200 series cards do not work with cuda 4.2 tasks under Linux!

Ok, thanks for the answer. It this case is there a way to have only cuda31 units ?
Cheers

Use an older driver, to prevent getting the CUDA4.2 app:
Uninstall the present driver completely. Find something pre-CUDA 4.2 (or whatever the dll's actually are). Probably around 265 to 285 should be good. Install these. Reset project and you should only get the 3.1app and thus 3.1tasks.

---

FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help