I believe that all SoCs, not just those in the M2 family, will take advantage of the optimizations. As an example, this optimization showing the potential of MetalRT.
Metal Viewport is actually still not complete but usable.I believe that all SoCs, not just those in the M2 family, will take advantage of the optimizations. As an example, this optimization showing the potential of MetalRT.
Blender Archive - developer.blender.org
developer.blender.org
By the way, it looks like Apple has gone back to Cycles optimizations after finishing the Metal backend for viewport.
Yes, this is nice for a semi-slim laptop. But i agree, we need something with about 512 apple gpu cores to compete with what’s on PC. A 7950x consumer grade rig with a good psu can host two 4090s. A big threadripper machine might host up to 4 dual slot blower cards (at a cost) Now, apple maybe doesn’t have to be that extreme to be usable, but to me it seems the new MP needs that quadra m2max at least.
Out of curiosity, what figures did you use for comparison? Techgage used the BMW and PartyTug scene to test some PC GPUs, but unfortunately the results are not comparable.
As expected they will go for Cycles optimizations. On the other hand, I believe ( looks like there is only me and myself ) Blender is not enough mature for long term projects now. Apple spends long time build up all system on a better way, with Metal UI optimizations first step passed before that UI performance was just below nightmare. Also they had to optimize Blender memory usage on AS macs. It is very suspicious now, large scenes or dense structures consumes too much memory from unified memory.I believe that all SoCs, not just those in the M2 family, will take advantage of the optimizations. As an example, this optimization showing the potential of MetalRT.
Blender Archive - developer.blender.org
By the way, it looks like Apple has gone back to Cycles optimizations after finishing the Metal backend for viewport.
What I saw on the mentioned video, if I m not wrong there was frequency dropdown on CPU side eventually overall performance dropdown. All I pointed that. Also I am very happy with this improvements and look forward M2 Ultra or what will Apple release beyond Ultra. M1 Ultra reaches kind of 'warm' behavior on render session just now, I assume if GPU frequencies will go higher M2 Studio Max/Ultra could be hotter.
Yes looking great so far considering I had a 14 core M1 Pro. In some cases it is close to 3x faster.
You don’t want to rely on that video channel, just to get an idea of the misinformation: https://forums.macrumors.com/thread...-channel.2378896/?post=31923135#post-31923135
I agreed about YT and misinformation, I just told what I saw on the screen not what he said on video.
maybe we will purchase e few of them this week but I tend to wait a few weeks more at least until March.
Yes, I was just about to post that nVidia cards use OptiX for their de-noising. If Apple can come with a Core ML denoiser then that would free up the GPU for more rendering.
Could a neural engine-based denoiser have more impact on rendering times than hardware-based ray tracing?
Denoiser and Optix are different pieces I guess. As far as I know, you could use them together, for still frames any Denoiser works more or less, for animation Optix accelerates whole rendering process + you can add Optix Denoiser for extra time savings, depends on which software or render engine you use. If you use any Nvidia HW with RT Optix could be default render engine option. However using any Denoiser for animation sequences not works for all scenes and ıt has got own risks. Most of the time not works as desired. It has different behaviors depending on which software you use.
No. Hardware RT will be huge.
Yeah Optix denoiser is mostly a separate piece of technology from the main Optix raytracing engine. The denoiser doesn't use the RT cores at all and just runs on the tensor cores (which are nVidia's version of the neural engine)