[SOLVED] Multitreading mantra errors again

   11752   17   2
User Avatar
Member
160 posts
Joined: Jan. 2006
Offline
Simple (or not simple) question.

What houdini build will not produce mantra multitread errors ?
I have strange errors with mantra rendering in versions from 182 to last build..

I start render, and go away, then i see red-colored mantra node and render is stopped in random time.

1. This error occur in random times. I`m do many tries and can`t understand the regularity rule..

2. This error occur on Win XP 64. Linux x86-32, Linux 86 OS. On Windows i see long long number of error code. On linux code number is “11”

3. Scenes for render contain the combinations of volumetrics, fluids or his caches, and lights..
Edited by - Aug. 19, 2008 09:07:53
User Avatar
Member
4140 posts
Joined: July 2005
Offline
I've not noticed this problem recently, but then I'm not rendering a lot of what you describe, that might be related. By way of a diagnostic, do you notice any difference if you render to disk? I always found more of these problems when rendering to mplay back when they happened.

Cheers,

J.C.
John Coldrick
User Avatar
Member
160 posts
Joined: Jan. 2006
Offline
Thanks, JColdrick, but when i render to disk, this problem is happened too

i have some idea:

How i can start render, and take some dump file ? For example, some flag for mantra or for houdini ? Some debug string…

With this option i can send dump file to SESI developers team, i think this is a best solution..

Can mantra create a some dump file ? Or, if it created, where i can find it on Linux and Windows systems ?

Thanks for answers…
User Avatar
Member
160 posts
Joined: Jan. 2006
Offline
Hm.. Right now i create some test scene, and start rendering. All ok..

i found some messages in system logs..

This very old logs:
Aug 5 02:27:02 livestation kernel: mantra-bin: segfault at b3e06008 ip 02b8dbf5 sp bfc33700 error 4 in libHoudiniUT.so
Aug 5 02:43:33 livestation kernel: mantra-bin: segfault at b7f07008 ip 01890bf5 sp b7e3aa30 error 4 in libHoudiniUT.so
Aug 5 17:29:16 livestation kernel: mantra-bin: segfault at b3cfa008 ip 08866bf5 sp bfe087a0 error 4 in libHoudiniUT.so
Aug 8 04:18:01 livestation kernel: mantra-bin: segfault at b3e02008 ip 02211bf5 sp b7e48e20 error 4 in libHoudiniUT.so
Aug 9 14:51:32 livestation kernel: mantra-bin: segfault at b3cbf008 ip 02bbebf5 sp bff69e20 error 4 in libHoudiniUT.so
Aug 9 15:14:34 livestation kernel: mantra-bin: segfault at b3e1e008 ip 0350ebf5 sp b7e3ce20 error 4 in libHoudiniUT.so


And this:
Aug 13 01:53:48 livestation kernel: mantra-bin: segfault at 31c1553140 ip 31c1553140 sp 44b9ea88 error 15

I think, libHoudiniUT.so - may be is a some files from GUI, may be mplay system files. But log from Aug 13 01:53:48 - i don`t have ideas.

Can i create dump files ?
User Avatar
Member
44 posts
Joined:
Offline
I'm getting the same errors also. It just stops rendering at at random points with the error code 11. I reported to SideFx about this last week. I'm running Ubuntu 8.04 64-bit.

At first I thought it was due to motion blur but I did tests with and without motion blur and it was still happening. Are you using the Particle Fluid Surface node? My scenes are using it and I think it may be the culprit. Hopefully, I'll hear from SESI soon since this is really stopping me from using Houdini.
User Avatar
Member
1002 posts
Joined: July 2005
Offline
If possible, please submit your test scene to support as this will help us track down any bugs that exist. Also provide details on your hardware, OS, and number of processors you're using for rendering and the frequency of the failure.

Andrew
User Avatar
Member
44 posts
Joined:
Offline
I have already done so when I reported the errors to SESI. Just taking a slight break from Houdini since these are errors are causing me headaches. I'm just glad to hear now that I'm not the only person seeing these errors. Luckily, the Olympics are here to keep me from going insane due to these problems.

I just had to stop since all I've been doing recently is bug hunt and not getting anything done.
User Avatar
Member
160 posts
Joined: Jan. 2006
Offline
andrewc, thanks, now i try to create lightweight simple scene, with this error receiving, and when i`m ready, i send bug report.

My scene now too big for email sending. I try to send via “Bug report” page, but receive error “file is too large” (i really like lock the some part of networks, and this produce large scene size, with locking for example, some volumetrics parts)

1. Now i`m disable all possible multitreading in all custom VOPs - no result. Errors again.

2. I don`t see mantra errors when only run mantra with -j1 flag, or set 1-processor rendering mode.


Today i want recreate some scene parts, and send little example + some system logs via “Bug report” sidefx.com page direct to developers.
User Avatar
Staff
181 posts
Joined: July 2005
Offline
If you have problems submitting the file, please email us directly: support@sidefx.com
User Avatar
Member
160 posts
Joined: Jan. 2006
Offline
Silvina thanks for answer,
I need some time for full report completion (this my home work-tests-demos only, not a job ).
User Avatar
Member
160 posts
Joined: Jan. 2006
Offline
completed.
User Avatar
Member
160 posts
Joined: Jan. 2006
Offline
wolf_cub_one:
Are you using the Particle Fluid Surface node?

No, i don`t use particle fluid surface.. Only volumes, and lights with depth shadow maps..

One question: Does you have volumetrics in the scene or not ? Or depth-map shadows ? If you work with scene without volumetrics, and it crashed with same “Error 11”.. then i`m don`t understand anything now..

I`m not sure now - i`m correct, or not with my declaration of this bug.
User Avatar
Member
44 posts
Joined:
Offline
I'm getting errors on the simplest scene. Not using Depth Maps or volumetrics but still get the code 11 errors and the renders stop. Not sure anymore what's causing the error. I'm using a quad core (Q6600) so maybe it is related to quad cores.
User Avatar
Member
160 posts
Joined: Jan. 2006
Offline
I'm using a quad core (Q6600) so maybe it is related to quad cores.

First i see this error on my Pentium D - dual core CPU. Yesterday i buy Quad Core 6700 CPU - and see error again. I don`t think what trouble is produced by kind of CPU..

Still waiting of developers team support
User Avatar
Member
160 posts
Joined: Jan. 2006
Offline
i`m afraid to say this..
(stop the breathing )
But i don`t see this “error 11” message anymore… In all my test scenes!

wolf_cub_one - try the last build..

Developers kung-fu in action!
:shock:
User Avatar
Member
44 posts
Joined:
Offline
Just done some tests and you're right it seems to be fixed!

Your file must have really helped them narrow the problem down. Thanks.

Silvina and Sidefx Tech Support, if you are reading this, thanks for the fix.

cheers
User Avatar
Member
5 posts
Joined: Sept. 2009
Offline
What build would that be?

I'm still receiving the segfault error (11) whenever I try to render anything with point clouds. To be honest, I've only tried it with Peter Quint's wetmap approach. When I hit render, it may render a few frames – the best I got were 25 in a row – but would also stop after the first frame.

I'm running:
- Ubuntu 10.04 64bit (besides the intended GCC4.4 build, I also tried the other ones, without luck)
- 4GB DDR2-800 RAM (after removing 2GB because they couldn't be run at full speed)
- NVIDIA GT9500 graphics card
- Houdini Apprentice 11.0.504 (that's not the latest build, right?)

Cheers!
- dr4ziw -
User Avatar
Member
5 posts
Joined: Sept. 2009
Offline
… could've been the .547 build.

Just downloaded and installed the just made available apprentice version.
Without wanting to get excited too early – 25 frames to go, at the moment –, it looks like it's working now.

Thanks a lot!

Cheers!
  • Quick Links