Can you please search the two variables using the search tool in the register editor to know if they are located in other folder?
If they are not set in your register, try to add them yourself in HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers
The variables must be DWORD (32 bits). In the register editor Edit->New->DWORD (32 bits) value
-Juanjo
lights make nvdia drivers crashes
Moderator: juanjgon
And if you use less lights it works? ... something is the scene is making the rendering really slow to crash the drivers. Do you have this problem with the IPR or in the final rendering?
If you want to send to me the scene I could test it here.
-Juanjo
If you want to send to me the scene I could test it here.
-Juanjo
i have rebooted of course... and i cant send the scene since i am on a LTE connection...
frankly this plugnis seems to me buggy when working with large files and heavy scene
.... the scene work if i erase most of the object layer so seems to be a problem of hugeness ,, unless some of the erased object layer had something wrong with it
frankly this plugnis seems to me buggy when working with large files and heavy scene
.... the scene work if i erase most of the object layer so seems to be a problem of hugeness ,, unless some of the erased object layer had something wrong with it
I have seen very large scenes render without problems. The drivers crashes are not due to large or complex scenes. The problem is that something in your scene makes the render really slow and the time needed to render each sample is too high.
I don't know why your system doesn't have the TdrLevel and TdrDelay variables in the register after install the Standalone, usually this help with your problem, and anyway if you set them yourself they should work.
What version of the plugin are you using, and what NVIDIA drivers? ... try to update you drivers if they are not the last available.
Can you please render your scene with the log enabled to see where the plugin is crashing and what CUDA error do you get?
I am sure that something is breaking the scene rendering. Usually the problem comes from something wrong in the kernel parameters, a lot of instances or hair in the sample place, a problem due to displacement mapping with large displacements, render very high resolution images with slow GPUs ... it is hard to say without the scene, but usually it is a problem that can be fixed.
The log file can help, please send it to me to know more about your issue.
-Juanjo
I don't know why your system doesn't have the TdrLevel and TdrDelay variables in the register after install the Standalone, usually this help with your problem, and anyway if you set them yourself they should work.
What version of the plugin are you using, and what NVIDIA drivers? ... try to update you drivers if they are not the last available.
Can you please render your scene with the log enabled to see where the plugin is crashing and what CUDA error do you get?
I am sure that something is breaking the scene rendering. Usually the problem comes from something wrong in the kernel parameters, a lot of instances or hair in the sample place, a problem due to displacement mapping with large displacements, render very high resolution images with slow GPUs ... it is hard to say without the scene, but usually it is a problem that can be fixed.
The log file can help, please send it to me to know more about your issue.
-Juanjo
here it is the log file.. nvidia drivers should had been updated when upgrading to win 10 and geoforce says they are updated
https://www.dropbox.com/s/64yovs6fdnec6 ... 9.log?dl=0
thanks
https://www.dropbox.com/s/64yovs6fdnec6 ... 9.log?dl=0
thanks
From the log file I can see that you are using a displacement map in one or more surfaces.
00:01:02 (0062.04) | <> Mem Used/Free/Total: 0 / 5034 / 6144
00:01:02 (0062.04) | <> Triangles/DisplaceTris/Hairs/Objects rendered: 2995966 / 783 / 0 / 267901
00:01:02 (0062.04) | <> Textures RGB32/RGB64/GREY8/GREY16 used: 0 / 0 / 0 / 0
Can you please disable the displacement to see if it is the source of the problems?
I can see also that your scene has 267901 objects, probably because you are using instances. A lot of instances in the same place, or instances with too big or too small scales can also crash the render kernel.
Anyway the scene seems to be rendered fine the first samples, and at some point it crash. It is a weird issue ....
The CUDA error is the 719
00:02:42 (0162.95) * OCTANE API MSG: CUDA error 719 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.
00:02:42 (0162.95) * OCTANE API MSG: -> kernel execution failed (dl)
00:02:42 (0162.95) * OCTANE API MSG: device 0: direct lighting failed
I will ask about this problem to the Octane developers, but anyway try to disable the displacement mapping and/or the instancing to try to isolate the source of the problem.
-Juanjo
00:01:02 (0062.04) | <> Mem Used/Free/Total: 0 / 5034 / 6144
00:01:02 (0062.04) | <> Triangles/DisplaceTris/Hairs/Objects rendered: 2995966 / 783 / 0 / 267901
00:01:02 (0062.04) | <> Textures RGB32/RGB64/GREY8/GREY16 used: 0 / 0 / 0 / 0
Can you please disable the displacement to see if it is the source of the problems?
I can see also that your scene has 267901 objects, probably because you are using instances. A lot of instances in the same place, or instances with too big or too small scales can also crash the render kernel.
Anyway the scene seems to be rendered fine the first samples, and at some point it crash. It is a weird issue ....
The CUDA error is the 719
00:02:42 (0162.95) * OCTANE API MSG: CUDA error 719 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.
00:02:42 (0162.95) * OCTANE API MSG: -> kernel execution failed (dl)
00:02:42 (0162.95) * OCTANE API MSG: device 0: direct lighting failed
I will ask about this problem to the Octane developers, but anyway try to disable the displacement mapping and/or the instancing to try to isolate the source of the problem.
-Juanjo
disabling displacement solved... but i wonder if the reason is because that way the scene is computational less heavy, or maybe there is incompatibility between using octane light and displacement at the same time?. i precise that in the area of scene where octane light where used there wasnt any materials with displacement
You can try to use the displacement with less resolution, if possible. Also if you can set you surfaces with less displacement distance, this could help.
The displacement mapping in Octane is computational heavy because It is computed at render time without fill GPU RAM with new geometry, something like a volumetric effect. Of course, if it is mixed with a lot of lights or other complex tasks for the render engine, the problem can be worse.
-Juanjo
The displacement mapping in Octane is computational heavy because It is computed at render time without fill GPU RAM with new geometry, something like a volumetric effect. Of course, if it is mixed with a lot of lights or other complex tasks for the render engine, the problem can be worse.
-Juanjo