Cuda error 999 ????? Help

[morphious2013]

Maybe it's because I built this scene in windows 7, and now opening it windows 8.1 pro? Also, Octane can't find material I used from Live DB. UGH. Here is the log, and also attached a pic of the error 999.

Started logging on 19.04.14 16:06:30

OctaneRender version 1.50 (1500001)

File Load Error: Could not load 'C:\Program Files\Smith Micro\Poser Pro 2014\B:rgba'.
File Load Error: Could not load 'C:\Program Files\Smith Micro\Poser Pro 2014\L:rgba'.
File Load Error: Could not load 'C:\Program Files\Smith Micro\Poser Pro 2014\T:rgba'.
File Load Error: Could not load 'C:\Program Files\Smith Micro\Poser Pro 2014\B:rgba'.
File Load Error: Could not load 'C:\Program Files\Smith Micro\Poser Pro 2014\H:rgba'.
CUDA error 999 on device 1: An unknown internal error occurred.
-> Failed to bind device to current thread
CUDA device 1: Failed to initialize resources
CUDA error 999 on device 1: An unknown internal error occurred.
-> Failed to bind device to current thread
CUDA device 1: Failed to initialize resources
CUDA error 999 on device 1: An unknown internal error occurred.
-> Failed to bind device to current thread
CUDA device 1: Failed to initialize resources
CUDA error 999 on device 1: An unknown internal error occurred.
-> Failed to bind device to current thread
CUDA device 1: Failed to initialize resources

[wimvdb]

Never seen the 999 error, can't help there
But regarding the missing textures there is a solution for the future:
In the setup of the plugin there is an entry for LiveDB texture map. By default it is set to a temp folder. When you moved from one OS to the other, that temp folder got lost. What I did is to change that path to a the texture folder in one of my runtimes (in my case P5\Runtime\Textures). I created an Octane folder in textures and that is what the path is set to.
Now that folder moves with the runtime (backup and move to a new machine). As long as you keep the drive letters identical octane will be able to find it

Paul has to answer that 999 problem

[morphious2013]

OK, sounds good. Thanks. Rebuilding the scene as we speak. When done, I'll post back what the results are.

[face_off]

A Poser scene with Octane node will contain texturemaps from your Runtime, and from your LiveDb texturemaps folder, and potentially other sources. When you move that scene to another PC, you need to make sure that Runtime has the same path on the new PC as the old PC, and you need to copy the LiveDb texturemaps (the folder which is specified in the plugin Configuration). If you built the scene with an older version of the plug (prior to where you could specify the LiveDb texturemap folder), the LiveDb texturemaps will be in the Poser temp folder from memory. If the plugin cannot find a texturemap it will ask for a new search path.

Cuda error 999 is a card failure error I think. Reboot your PC. Make sure you have not overclocked your card, and that you are not rendering in OctaneRender for Poser AND OctaneRender Standalone at the same time.

Paul

[morphious2013]

Yea, that seems to be what happened. When I upgraded to Windows 8.1, it has a different file structure and could not find the original location. Rebuilt the scene in New setup, and so far so good. Thanks.

[Sinuhe]

Never seen the 999 error, can't help there

Hi, I see lucky man there is .
Since installing version 1.51 and plug-in 1.50 CUDA error 999 is my everyday and nightmare. I see window like this in every 10 to 30 minutes of rendering, since one month I cannot do anything serious .
Some datas:
Poser Pro 2014 SR 3.1
Octane plug-in 1.50
Standalone vers. 1.54
Win 7 Home Premium with all corrections
GForce 580 GTX @ 1,5 GB VRAM with actual driver, without overclocking any other "modifications"
Temperature Vprocessor 73-74 C stable



The situation is with old files and the same with new files. 999 error is breaking any rendering. I do not use any Live DB materials, only my own from HDD or saved on HDD. I uninstalled previous versions and installed present versions without any software working in background, I do not render with standalone version in the same time and I didn't change my installation system for last 3 or 4 years. So situation is very clear and stable

Can someone help me? To be honest I've get enough

[face_off]

Hi - I'm sorry, but this looks like a video card hardware problem to me.

I see the problem occurs on PMC - do you also get it on Direct Lighting and Path Tracing?

Does the card sit at 73 degrees - or get hotter? Perhaps it has been cooked.

Do you have another card you can try?

If you can get the render started, export to ORBX, then load into Octane Standalone and render from there to see if it pops the same error.

Have you tried UNDERCLOCKING the card a little (this has worked for a number of people with older cards)?

Paul

[Sinuhe]

Hi Paul. Forgive me I muted for last days, however I really appreciate Your hints and advices .
1. I can't try another vcard, I am to poor for this I would rather buy new computer than only card
2. I tried to check all other possibilities and what is my fault, I miss latest drivers update for nVidia from 05-26-2014. Mea maxima culpa
3. However I have no error 999 for now, my new hobby is error 700. In the same pure shape mentioned by RGUS as follows: " CUDA error 700 on device 0: An exception occurred on the device while executing a kernel. The context cannot be used anymore and must be destroyed. All existing device memory allocations from this context are invalid and must be reconstructed."
4. I tried with imported file to standalone version 1.54 - the same with all kernels. Of course it makes me off any rendering. I check it all with newly created file in Poser.
5. Max. temperature at my vcard proc. is stable 73-75 C, I tried test my card with last year Tomb Rider (I know it s weird way to test ) and the game with normal accompaniment of groans of funs was playable and didn't hang or something like this.
6. Today I plan to go back to versions 1.20 and watch what will happen

Greetings Jacek

[face_off]

Did you try my suggestion above of underclocking the card?

[Sinuhe]

Hi again. Just finished with my trials and changes. So, I underclocked my vcard according to Your suggestion. About 40 % down was stabilized the situation, however rendertime dramatically increased ab. 3-4 times, depending of used kernel. Moderate complex scene, which was rendered 3-4 hours in this case was indicated as 9-12 hours rendertime. I am old man and I have not enough time for this, so I went back to versions 1.20. All was back and is stable as few months ago.
To be honest I regret all new options from 1.5 version, but I don't see any reasonable way to solve this problem, except buying strong machine . In my humble opinion nVidia GTX 580 is not enough for Octane 1.50 yet, that's all. Thank You Paul for Your help and assistance
Cheers Jacek