Tweet
Hi!
I am using VRay from Grasshopper to create animations. The CPU engine has been working fine for me, but it was a bit slow, and I have a GPU card idling out there. But when I switched to the CUDA backend, the rendering was always interrupted by the following error:
Based on the suggestion in this post on the same issue, I thought it was my VRAM is not large enough. I created a minimal working example (Grasshopper script attached below) on a scene with just a simple box and a plane, but the error kept showing up when rendering in animation + CUDA mode.
I switched back to single-frame rendering mode with the CUDA engine. I clicked `render`, and it was interrupted again by the error message above. Following the "try-it-again" suggestion in this post, I made no change and pressed `render` again in the CUDA engine, and suddenly it can render (and much faster than CPU as expected)!
In summary, the CUDA engine must be tried twice (once failed with the GPU light cache error above and once succeeded) to work on a small example. If this is always the case, from a user's perspective, can we simply have the VRay engine automatically try it 1-5 times until it works when using the CUDA engine?
The grasshopper file to reproduce the issue is here (why aren't we allowed to upload file with an extension of `.gh` in this forum???).
My system info:
This is my GPU system info from the Nvidia control panel:
Full Vray log can be found here. vray_log.txt
Thank you very much for your help in advance!
I am using VRay from Grasshopper to create animations. The CPU engine has been working fine for me, but it was a bit slow, and I have a GPU card idling out there. But when I switched to the CUDA backend, the rendering was always interrupted by the following error:
Error : [GPU light cache] Could not allocate buffer of size 1865MB!
Based on the suggestion in this post on the same issue, I thought it was my VRAM is not large enough. I created a minimal working example (Grasshopper script attached below) on a scene with just a simple box and a plane, but the error kept showing up when rendering in animation + CUDA mode.
I switched back to single-frame rendering mode with the CUDA engine. I clicked `render`, and it was interrupted again by the error message above. Following the "try-it-again" suggestion in this post, I made no change and pressed `render` again in the CUDA engine, and suddenly it can render (and much faster than CPU as expected)!
In summary, the CUDA engine must be tried twice (once failed with the GPU light cache error above and once succeeded) to work on a small example. If this is always the case, from a user's perspective, can we simply have the VRay engine automatically try it 1-5 times until it works when using the CUDA engine?
The grasshopper file to reproduce the issue is here (why aren't we allowed to upload file with an extension of `.gh` in this forum???).
My system info:
HTML Code:
VRay for Rhino 7 - 6.00.01 - Nov 3 2022 V-Ray COre - 6.00.21 V-Ray NeUI - 5.0.1 V-Ray AppSDK - 4.00.00 Windows 10, 64bit CPU Intel(R) Core(TM) i7-6820HQ CPU @ 2.70GHz 2.70 GHz Quadrop M1000M, Driver 527.27
HTML Code:
[Display] Operating System: Windows 10 Pro, 64-bit DirectX version: 12.0 GPU processor: Quadro K620 Driver version: 496.13 Driver Type: DCH Direct3D feature level: 11_0 CUDA Cores: 384 Core clock: 1058 MHz Memory data rate: 1.80 Gbps Memory interface: 128-bit Memory bandwidth: 28.80 GB/s Total available graphics memory: 18377 MB Dedicated video memory: 2048 MB DDR3 System video memory: 0 MB Shared system memory: 16329 MB Video BIOS version: 82.07.B3.00.01 IRQ: Not used Bus: PCI Express x16 Gen2
Thank you very much for your help in advance!
Comment