Author Topic: Some questions about vlar kill (Read 20861 times)

diederiks · « **on:** 06 May 2009, 10:54:23 am »

Since i only use my gt8500 for seti@home WU crunching with v10 of the optimized seti@home cgpu client, is see loads of vlar kills in the crunched WU, why is this done?
And is there a optimized client that crunches vlar WU?
And what's the impact of crunching these WU on a video cart.
I ask thise becaus my video cart is often doing nothing because of the vlar kill.

Raistmer · « **Reply #1 on:** 06 May 2009, 12:27:08 pm »

If your GPU is idle then better to use re-schedule script time to time to feed it. Processing VLARs on GPU is very ineffective still.

M_M · « **Reply #2 on:** 07 May 2009, 03:32:48 pm »

Where can we find this script? What exactly does the script do? How to use the script?

Please explain a bit for all of us dummies...

Thanks in advance.

Raistmer · « **Reply #3 on:** 07 May 2009, 04:06:17 pm »

http://lunatics.kwsn.net/12-gpu-crunching/cpu-gpu-rebranding-perl-script.0.html

BMaytum · « **Reply #4 on:** 27 May 2009, 12:22:53 pm »

@ Raistmer or Richard Haselgrove:

After irrecoverable Stop 0x051 Registry BSODs in Feb09, I've reverted to a Jan2008 backup and installed/re-installed many updates & programs. Recently (as per Raistmer's guidance http://lunatics.kwsn.net/12-gpu-crunching/v10-of-modified-seti-mb-cuda-opt-ap-package-for-full-multi-gpucpu-use.msg17425.html#msg17425, I have:
1) Installed nVidia v185.85 drivers;
2) Clean Install BOINC 6.6.28 from scratch (Jan08 backup had no BOINC or S@H installed);
3) Used Jason Gee's Beta unified optimized S@H Installer;
4) Got the v2.2 cufft.dll & cudart.dll in the appropriate (default) folders.

However, I've had numerous MainBeam WU's terminate with computational error, e.g.:

stdoutdae.txt:
27-May-2009 07:30:18 [SETI@home] Starting 03fe09ad.15841.1299.10.8.90_0
27-May-2009 07:30:18 [SETI@home] Starting task 03fe09ad.15841.1299.10.8.90_0 using setiathome_enhanced version 608
27-May-2009 07:30:19 [SETI@home] Computation for task 03fe09ad.15841.1299.10.8.90_0 finished
27-May-2009 07:30:19 [SETI@home] Output file 03fe09ad.15841.1299.10.8.90_0_0 for task 03fe09ad.15841.1299.10.8.90_0 absent

And stderr out:
<core_client_version>6.6.28</core_client_version>
<![CDATA[
<message>
- exit code -1073741511 (0xc0000139)
</message>
]]>

Validate state Invalid
Claimed credit 0
Granted credit 0
application version 6.08

You can see more details for my Host here: http://setiathome.berkeley.edu/show_host_detail.php?hostid=4950990

Questions:
1) Does exit code -1073741511 (0xc0000139) just mean these WUs were terminated by VLARKill? or, is there something wrong with my setup/ config/ project preferences? (My app_info.xml is not modified from what JGee's Beta Installer created).
2) Today I've gotten no Cuda WUs - is that because I've exceeded some #error limit(s)?

Raistmer · « **Reply #5 on:** 27 May 2009, 12:43:33 pm »

This your result because of VLAR-kill:
http://setiathome.berkeley.edu/result.php?resultid=1239075202

<core_client_version>6.6.28</core_client_version>
<![CDATA[
<message>
- exit code -6 (0xfffffffa)
</message>
<stderr_txt>

VLAR WU (AR: 0.011061 )detected... autokill initialised
SETI@home error -6 Bad workunit header

File: ..\worker.cpp
Line: 144

</stderr_txt>
]]>

All other errors - not.

BMaytum · « **Reply #6 on:** 27 May 2009, 01:09:44 pm »

Thanks for reply Raistmer. I now understand that exit Code -6 results from Vlar Kill code.

Admittedly, when I saw a rapid series of Cuda WU Compute Error terminations over the past 2 days, I've tried various combinations of cudart.dll and cudafft.dll versions (v2.2 mixed with earlier versions) thinking perhaps there was some incompatabilities, but I didn't keep track of those dll version combinations relative to which Cuda WUs completed and which Wus terminated with errors.

I've verified that Currently I've using v2.2 of both dll's but since I have no Cuda WUs in cache I'll have to wait & see whether I get exit code -1073741511 (0xc0000139) again. In the meantime can anyone provide insight & fix for exit code -1073741511 (0xc0000139)?? I've searched Lunatics and S@H "Number Crunching) forums for that code but found no posts for it.

Raistmer · « **Reply #7 on:** 27 May 2009, 01:13:30 pm »

link on result with this code?

Jason G · « **Reply #8 on:** 27 May 2009, 01:20:05 pm »

http://www.boinc-wiki.info/Unrecoverable_error_for_result_%27(result)%27_(_-_exit_code_-%27(number)%27_(%27(hex-number)%27))

[Edit: Fixed link to point to top of article page.]

BMaytum · « **Reply #9 on:** 27 May 2009, 01:42:19 pm »

Quote from: Raistmer on 27 May 2009, 01:13:30 pm

link on result with this code?

@ Raistmer: Here's several Cuda WUs that got exit code -1073741511 :
http://setiathome.berkeley.edu/result.php?resultid=1239045099
http://setiathome.berkeley.edu/result.php?resultid=1239045070

@ Jason G: Thanks for the Boinc Wiki link, but it doesn't list exit code -1073741511 (0xc0000139), did I miss it (or have I "discovered" or created a new one)? Just a guess: my error code is not a Boinc error, rather it's an S@H app error (or a result of me possibly mixing incompatible DLLs/EXEs together.

[Edited my text @ Jason G]

Jason G · « **Reply #10 on:** 27 May 2009, 01:58:29 pm »

Quote from: BMaytum on 27 May 2009, 01:42:19 pm

@ Jason G: Thanks for the Wiki link, but it doesn't list exit code -1073741511 (0xc0000139), did I miss it (or have I "discovered" or created a new one)?

Hi,
The reason to post that link was not to find a specific code number: but to see that 'those kind' of errors (basically Windows stop errors) are generally indicative examples of a combination of 'weirdness', usually drivers or hardware, or some similar combination of factors. The specific part of the list is a brief & basic rundown common known occurrences at the time it was written, that the type of error are not limited to the list.

One example that is in the list, Exit Code -1073741819 (0xc0000005), I had occur on my old p4 back when I started, a machine which had no connection whatsoever to Climate prediction as described (Since never attached to that project), but happened to have an ATI radeon card that had been suffering from a rash of useless driver revisions. As soon as that particular issue was resolved several driver revisions later, then the error never reoccured.

So there Is certainly some issue, and it could be anywhere, the task is to isolate it methodically. If you need some general guidance on how to do so, then I can lend a hand on the weekend.

Jason

BMaytum · « **Reply #11 on:** 27 May 2009, 02:14:42 pm »

Jason G:

OK now I understand the intent of your earlier post, thanks for the followup clarification to my overly literal mis-interpretation.

I suppose it's possible there's weirdness somewhere on my PC e.g. nVidia 185.85 driver installation (I simply ran the driver installer w/o first uninstalling prior driver(s), etc. Games work fine but I know that's not primafacie 'validation' that all/anything is correct.

Anyway, once I get & crunch some new Cuda WUs I'll decide if I need your guidance to methodically track down the cause if the exit code -1073741511 (0xc0000139) recurs. Thanks for that offer and your time today (+ your Beta Installer!).

Jason G · « **Reply #12 on:** 27 May 2009, 02:21:44 pm »

Cheers,
About that Video card driver: Just keep in the back of your mind, compared to what has been before, they're pretty good (well I had some pretty crappy nVidia drivers anyway

). I'm not saying it couldn't be something with that, but nothing about that error specifically says that there is or isn't either. I would be looking at things like mobo chipset drivers, running memtest86+ for at least a few passes, pulling equipment out part by part until the problem goes away, things like that. [Edit: An artefact scan, using 'AtiTool' , might help reduce or confirm suspicion from the video end of things]

BMaytum · « **Reply #13 on:** 31 May 2009, 02:18:33 pm »

@ Jason G: "ATiTool" is N/A since I'm nVidia-based (Vidcard & Mobo).

With apologies to Raistmer since my postings have gone off-topic from Vlar Kill per-se, here's an update on my prior Exit Code -1073741511 (0xc0000139) concerns. (Again, I know as of 27May & later I solely used v2.2 from Raistmer's link in the now-locked V10/V11 thread but I may have had various mixture(s) of cudart.dll & cudfft.dll versions prior to that).

Per http://setiathome.berkeley.edu/results.php?userid=7782289&offset=20&show_names=0&state=5 these
MB WUs terminated on Host ID http://setiathome.berkeley.edu/show_host_detail.php?hostid=4950990
with Exit Status -1073741511 (0xffffffffc0000139):

WU Name / Task ID / Workunit ID
28, 29, 30, 31 May 2009 NONE!!

sunu · « **Reply #14 on:** 31 May 2009, 05:26:32 pm »

Quote from: BMaytum on 31 May 2009, 02:18:33 pm

@ Jason G: "ATiTool" is N/A since I'm nVidia-based (Vidcard & Mobo).

Despite their name, ATI tray tools can also be used with nvidia cards.

Author Topic: Some questions about vlar kill (Read 20861 times)

diederiks

Some questions about vlar kill

Raistmer

Re: Some questions about vlar kill

M_M

Re: Some questions about vlar kill

Raistmer

Re: Some questions about vlar kill

BMaytum

Re: Some questions about vlar kill

Raistmer

Re: Some questions about vlar kill

BMaytum

Re: Some questions about vlar kill

Raistmer

Re: Some questions about vlar kill

Jason G

Re: Some questions about vlar kill

BMaytum

Re: Some questions about vlar kill

Jason G

Re: Some questions about vlar kill

BMaytum

Re: Some questions about vlar kill

Jason G

Re: Some questions about vlar kill

BMaytum

Re: Some questions about vlar kill

sunu

Re: Some questions about vlar kill