Advertisement

You're blocking ads, which pay for BlenderNation. Read about other ways to support us.

Blender Cycles: OpenCL now is on par with CUDA

10

AMD videocard owners rejoice! With the work on the split Cycles OpenCL Kernel, the performance of AMD GPU's has increased dramatically.<!--more-->

Cycles has a split OpenCL kernel since Blender release 2.75. It's an alternative approach to what is used on CPU (so called megakernel). The idea behind splitting the kernel is to have multiple smaller kernels which are not only simpler and faster to compile, but also have better performance. Initial split kernel patch was done by AMD. Further work was also funded by AMD.

The OpenCL Split Kernel now supports nearly all features. Only Correlated multi jitter is missing. Baking works, but uses the mega kernel. Volumetrics, SSS, Branched path tracing, HDR lightning and Denoising are fully supported.

With current drivers, all production files from the official cycles benchmark pack, including the huge one from Gooseberry, render now pretty fast.

About the Author

Avatar image for Bart Veldhuizen
Bart Veldhuizen

I have a LONG history with Blender - I wrote some of the earliest Blender tutorials, worked for Not a Number and helped run the crowdfunding campaign that open sourced Blender (the first one on the internet!). I founded BlenderNation in 2006 and have been editing it every single day since then ;-) I also run the Blender Artists forum and I'm Head of Community at Sketchfab.

10 Comments

  1. nice to see people aren't going to be forced into getting Nvidia cards anymore.
    except for the lack of a high end AMD card, that comparison seems to indicate that "on par" was being nice to Nvidia.

    i wonder how 2 or 3 rx480s would stack up against a single 1080 or 1080ti respectively.

  2. Nice, but since Cuda is standard not only in blender but in many others software I'll stick with it .
    However good to see so many improvements from amd lately, I hope this will keep competition high

    • Technically, to nitpick, you could say that OpenCL is not only in nVidia GPUs but also in other hardware, so it would be better to stick with it ;)

      (Yes, I am really not into vendor lock-in, you would have guessed)

      I am really thrilled to see such improvements! This also makes me wonder... Did anyone try the OpenCL backend on a FPGA?

  3. Even as somebody who uses 5 different nVidia cards across three PCs, it's good to see Blender helping AMD users feel some warmth.

  4. I have a 1080 card at work, and it gives cuda "ctx" errors attempting to render anything more complex than extremely simple scenes. I so badly want to try out gpu rendering, but it seems to never work, even on substantial hardware :(

    • @Steve
      i've got a titan X (maxwell) at work and a rx480 at home.
      both have been working fine for me though the slow loading of the amd split kernel can be a little annoying at times.

      the only problem i ever had with gpu rendering was when i had a card that only had 2gb vram and was trying to render a complex scene. just saying that it seems to pretty much always work in my usage because it leads me to believe there may be something else going on in your setup or maybe there are still some problems with the nvidia pascal line in blender.

Leave A Reply

To add a profile picture to your message, register your email address with Gravatar.com. To protect your email address, create an account on BlenderNation and log in when posting a message.

Advertisement

×