Announcement

Collapse
No announcement yet.

What hardware would I need to speed up "Building Light Cache"?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • What hardware would I need to speed up "Building Light Cache"?

    Ok so I have an Intel 8700k 6-core. Its a great CPU but there are Xeons and Threadrippers with many more cores. I am also using a GTX 1060 just for my display and an RTX 2070 just for rendering.

    It's clear to me that a better GPU like a 2080ti will give me faster renderings, but the light cache and image save to file times don't utilize the GPU. Is there any way to change this? It seems weird having to upgrade from an 8700k, but I hate waiting for there light cache. It's especially annoying whenever I make tweaks to test a change with a low resolution. I will click the render button hundreds of times just to see quick previews, but no matter the resolution, the cache times are always the same.

    If I am stuck with CPU, is there a limit the core count that will speed up the cache?

    Thanks guy!

  • #2
    Hi akashi1

    Please consider using Brute Force + Brute Force for Global Illumination as a solution.
    That way Light Cache will be disabled and the CPU will not be used for the calculation of the GI.
    Output saving times are dependent on CPU and Hard Drive speed only, currently there is no way to engage the GPU in this process. Additionally, please note that if you are saving the output file on a network location - the speed of your network can also influence the time it takes to save the image.
    Furthermore, for quick previews you could use the Interactive render option (V-Ray Asset Editor > Settings > Render > Interactive) and instantly see the changes you make on the scene.
    And finally,V-Ray is able to take full advantage of the CPU, which means that you will be able to utilize all of your CPU cores for calculating GI and rendering.
    Last edited by slavcho.brusev; 13-04-2020, 02:52 AM.

    Comment


    • #3
      akashi1 My experience is that the LC calculation is quite fast, so maybe your LC setup can be adjusted for more speed. My basic setup is 1000 subdivs, 0,01 size and retrace 1. For large images I use 2000...3000 subdivs. The calculation needs on a 16-core CPU only a few seconds most. For scene setups the interactive mode is a very nice option and the LC isn't used. I'm not sure, but for the final pass the LC could help to get a faster rendering since the glossy effects should be calculated faster per LC.
      www.simulacrum.de ... visualization for designer and architects

      Comment

      Working...
      X