Blender 3.0's Cycles X Rendering Performance Is Looking Great

[phoronix]

Phoronix: Blender 3.0's Cycles X Rendering Performance Is Looking Great

A status update on Blender's "Cycles X" project was published today ahead of the upcoming Blender 3.0 release and with already some feature additions planned for Blender 3.1...

https://www.phoronix.com/scan.php?page=news_item&px=Blender-3.0-Cycles-X-State

[chuckula]

Instead of having to have a different (complex) backend for every type of GPU it would be nice if there was a more uniform & open source API that could work on all the major GPUs as long as they have the right hardware....

[waxhead]

Instead of having to have a different (complex) backend for every type of GPU it would be nice if there was a more uniform & open source API that could work on all the major GPUs as long as they have the right hardware....

So you want OpenCL back then

[agaman]

If Apple can add a Metal backend for Cycles X, would not be possible to add a Vulkan compute backend too?

[amxfonseca]

If Apple can add a Metal backend for Cycles X, would not be possible to add a Vulkan compute backend too?

For sure. Someone needs to sponsor and develop it. I don't think either Apple or Nvidia are going to do it, since there is no monetary incentive for them. And AMD seems to be investing on the HIP backend already, so it would be also wasteful for them, specially if they can't extract the required performance from it, last thing you want to is to develop a backend that will run better on your competitors hardware.

So any company that sells an high performance device that supports Vulkan can definitely support the development of a new backend.

[microcode]

Instead of having to have a different (complex) backend for every type of GPU it would be nice if there was a more uniform & open source API that could work on all the major GPUs as long as they have the right hardware....

OpenCL never really got implemented consistently well, but Vulkan Compute is stable, performs well on every vendor. It is a little bit of a different programming model. The other nice thing about using Vulkan is that they can integrate it in hybrid situations with Eevee and the viewport renderer, as they move the viewport and Eevee to Vulkan.

The CUDA backend is courtesy of NVIDIA money, so no sense being sour about it; but it would be nice to see somebody put up for the Vulkan backend (maybe Epic + AMD, and Epic could lend something better than money: expertise).

[rmfx]

So you want OpenCL back then

No, he wants HIP. HIP is designed to work with NVidia hardware too. They just need to complete it and make it available for all devices.

Can Vulkan compute do the same as HIP/opencl/CUDA or it has strong limitations? If that was so great, why HIP would even be made ?

[ddriver]

For an open source project, it sure has a boner for using proprietary tech.

"why HIP would even be made ?"

One word - beads!

[tildearrow]

So you want OpenCL back then

You all forgot about Vulkan Compute.

[stargeizer]

Coders sure choose CUDA because when you make a struct FooBar{}; in CUDA, it works on both CPU-side and GPU-side, no pains, no complications, easy peasy.

Vulkan / OpenCL / anything not CUDA, don't have any data-structure sharing like that with the host code. Its a point of contention that makes anything more complicated than 3-dimensional arrays hard to share, requiring lots of workarounds, and you need to consider what are the hard limits of GPU to use.

Yeah, Vulkan / OpenCL have all sorts of pointer-sharing arrangements (Shared Virtual Memory), but its difficult to use in practice, because they keep the concepts of "GPU" code separate from "CPU" code. CUDA doesn't have that weakness, so there's no limitations of what you can do in a GPU, code can be battletested faster, maintenance is lightyears easier and people can go to their homes earlier in the night.

That's why nVIDIA won the compute wars years ago, and that's why other manufacturers are years behind of work to just play catch-up, and given the actual state of Vulkan tooling, nvidia has no reason to worry about. And Apple doesn't want to play this tune again, that's for sure. (Honestly, i was hoping Intel One API was a game changer, but is focused more on the datacenter than anything else).