Have the patents Img Tech have on tile based deferred renderers expired?Apple penned a new IP deal with them last year. Even at the time it was speculated to be about ray tracing:
Apple’s Imagination Technologies deal is all about ray tracing and AR
Apple and GPU designer Imagination Technologies forged a mysterious new multi-year agreement, almost certainly to bring ray tracing to Apple devices.venturebeat.com
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Ray tracing on Apple Silicon?
- Thread starter sunny5
- Start date
- Sort by reaction score
Don’t think so but prior to this Apple was no longer licensing anything substantial from, and was in dispute with, ImgTech for several years. This was seen as a big expansion of Apple’s interest in ImgTech’s IP rather than any kind of settlement. Nobody really knows and both sides are being coy about exactly why Apple penned a new deal - with ImgTech saying only that “It’s an expanded license, a wide range of IP, for multiple years. I’m sure you can speculate. Everyone is speculating. But we can’t say.”
Imagination Technologies charts its future with new Apple deal and post-MIPS strategy
Imagination Technologies surprised many by announcing on January 2 that it had again signed a multi-year licensing deal with Apple.
If that’s accurate then it sounds to be about more than just TBDR and the biggest new thing ImgTech have developed is ray tracing. So still speculation, but not unfounded.
Dissecting the Apple M1 GPU, part I | Hacker News
news.ycombinator.com
It seems a lot of IMG’s areas of interest (https://www.imaginationtech.com/) align with Apple’s.
Raytracing, AI, VR etc (GPU, Car and VR/AR headsets)
Is Keyshot Ready for the Apple M1 MAX Chip?
Will there be a small update or will it come with the release of Keyshot 11?
"Unfortunately the KeyShot 11 release will not support GPU rendering on Mac Operating Systems.
One of the many reasons is due to Apple GPUs not possessing RayTracing cores, a key feature in KeyShot's programming.
While I know this is not the answer you were looking for, I want to assure you that things are in happening behind the scenes to better support KeyShot on Mac OS.
Med Venlig Hilsen / Best Regards,
Erik Williams
Customer Support Specialist"
It seems like we will have to wait for apple silicon to get RT cores for some programs like Keyshot to get GPU support. Not sure about a native cpu only keyshot release though.
Wonder if Toolbag 4 is waiting for something similar before bringing over native support.
That does sounds like a weird marketing-formulated excuse. Presence or absence of RT cores only affects performance. From the feature and programmability standpoint M1 might as well contain RT cores…
P.S. I had a quick look and Keyshot does support GPUs like Nvidia P1000 that do not have RT cores either. So yeah, just an excuse.
Last edited:
Yeah I thought it was an excuse too. Think i mentioned that in the other 3D thread. Still I think it basically means we have to wait until apple silicon has rt cores for them to consider adding gpu acceleration.
I tried the demo earlier today and the cpu only via rosetta made my m1 struggle hard. At least on marmoset toolbag 4 i can move around and use the program. It even supports ray tracing.
Blender Archive - developer.blender.org
MetalRT closely follows the Optix implementation, and in some cases (notably handling of transforms) it makes sense to extend Optix special-casing to MetalRT. For these generalisations we now have KERNEL_GPU_RAYTRACING instead of KERNEL_OPTIX.
- MetalT92212 4: Would like to move quickly to get a fully working version in the code base as quickly as possible. Currently only 4 test cases are failing. We would also like to get the Metal RT patch in. Looking into using run time compilation to specialise the rendering pipeline which had 5%-10% performance impact. However, compilation can take a long time negating the benefit. We are wondering if Blender is set up to build and test Metal builds. Finally, the Metal implementation currently does not have an lgamma implementation that is usable for licensing reasons and this is currently used by some closures in Cycles so one needs to be found that can be used. D13263 3, D13241 3, D13243 3, D13236 3. (Micheal)
- Currently Blender can easily support Metal builds as it currently runs on a M1 laptop so it should easily be able to support testing Metal. (Brecht)
Do you mean adding ray tracing cores to the GPU like RTX or do you mean bundling ray tracing in its own accelerator on the SOC a la the NPU?
The how and when is what we’re discussing. Apple is obviously setting up for the addition of such capabilities with software APIs but the reason they haven’t shipped anything yet is that Apple didn’t have (as far as we know) any hardware ray tracing IP to do so. It’s likely that’s what their recent deal with ImgTech was to gain access to. If Apple’s recent moves that @hefeglass linked to with blender are anything to go by, we’re likely to see hardware accelerated ray tracing in Macs sooner rather than later.
Last edited:
Could Apple's hardware ray tracing be similar to this? https://imaginationtech.com/products/gpu/img-cxt-gpu/
Which device could get it first? iPhone? Mac Pro? MacBook Pro?
If it is separate silicon (like nvidia) I would figure the desktop/notebooks will see it first as the phones don’t really have the space to spare.
i could be wrong though. ??♂️
That’s what the smart money says Apple’s recent IP deal with them was for. As for when it’ll appear and on what device first … ?♂️ I could make arguments for any device level. I think soon though.
Its not a straightforward thing to do. But as others have said, Apple is most certainly working on it and they have recently bought IP from IMG. Metal is RT ready.
RT hardware has to be integrated into the GPU, I just don’t see how it can be a separate coprocessor, communication overhead will make everything incredibly slow. Bounding box acceleration is trivial, the tricky part is optimizing memory access. The papers I’ve seen trying to figure out how Nvidia does it seem to suggest that “RT cores” are closely integrated with the texture units and their main job is to batch memory accesses to improve cache coherency.
When could Apple be prepared to ship ray tracing hardware if it uses Imagination IP? This fall? Next year?
Which devices could Apple choose to have ray tracing hardware? Only computers? iPad? iPhone?
Could Apple ship ray tracing hardware only in Pro devices?
Blocks of the Imagination's GPU with ray tracing.
The main components of the GPU include:
• Unified Shading Cluster (USC) – the compute heart of the GPU, a multi-threaded programmable SIMT processor which can simultaneously process pixel data, geometry data, compute data as well as 2D/copy housekeeping tasks. More USCs equates to higher compute performance for the GPU configuration.
• Texture Processing Unit (TPU) – handles texture addressing, sampling and filtering in highlyoptimised logic. More texture unit equates to higher visual complexity, greater refresh rates and increased display resolution support.
• Raster/Geometry Block – a collection of fixed-function units enabling post- and pre-processing of data before/after processing by the USC including culling, clipping, tiling, compression, decompression, iteration, etc.
• Top-level (CXT RT3) – including L3 cache, AXI bus interfaces and firmware processor
• Ray Acceleration Cluster (RAC) – a new dedicated block for efficient handling of all ray tracing processing stages.
Source: Imagination's white paper "Rays your game" https://imaginationtech.com/resources/rays-your-game/
Last edited:
For Apple? Rendering apps for quick previews (since from my understanding everyone does final product in CPU). I doubt they will tout games as the best use if it.
Not a separate coprocessor. A dedicated block in the GPU itself. AMD doesn't have a dedicated ray accelerator. They "reuse" the TMU. Nvidia doesn't "reuse" or dual purpose any of it's ray acceleration hardware.
I would advise care when looking at block diagrams like that. It’s just a logical partitioning, the hardware does not have to look like that.
Metal RT is kind of geared to do production rendering acceleration on the GPU. And regarding gaming - sure, why not? Apple already has tutorials on how to do hybrid RT shadows or ambient occlusion in games.
And how do you know that? Because Nvidia gives it a different name? We don’t have any internal documentation on how the RT cores work. Why would they be a separate block? Making them a part of TMU makes a lot of sense, because it’s the part that has to deal with memory locality. And locality of access is the biggest issue for GPU-driven RT.
P.S. This is the paper I am referring to: http://ceur-ws.org/Vol-2485/paper3.pdf
Last edited:
They do, and so far no one has seemed to take them up on the offer (the one game they touted during last years WWDC [Metro Exodus] didn't bother to even add those effects in the basic version of the engine).
Because Nvidia says they are a separate block/core? ??♂️
Page 12 has figure 3 which shows the separate texture units (4 per sm) and the RT cores below it.Page 10 https://images.nvidia.com/aem-dam/en-zz/Solutions/geforce/ampere/pdf/NVIDIA-ampere-GA102-GPU-Architecture-Whitepaper-V1.pdf said:
It certainly seems like Apple’s relationship with Imagination has turned out to be a fairly smart bet, as I do not see Adreno or Mali architectures from the likes of Qualcomm or ARM to have these capabilities. I heard Samsung are trying to adapt some of AMD’s graphics technology for use in low-power devices but that doesn’t seem to be a simple process.
What jumped out at me from the Photon pages on Imagination’s website were the specs: 7.2 GRays total, 9 TFlops (32-bit) performance, aimed at a mobile processor. That’s no small amount compared to the M1’s current totals. It also seems like this is a whole new graphics technology with 50% greater geometry and shading throughput compared to previous generations.
What jumped out at me from the Photon pages on Imagination’s website were the specs: 7.2 GRays total, 9 TFlops (32-bit) performance, aimed at a mobile processor. That’s no small amount compared to the M1’s current totals. It also seems like this is a whole new graphics technology with 50% greater geometry and shading throughput compared to previous generations.
The crazy thing to me is no one else is using IMGTec stuff but Apple. For it to be as good as it is, with no one else using it is weird.