• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Sony files new patent for ‘Accelerated Ray Tracing’ - speculation that it could be more performant

WitchHunter

Member
Source: https://www.psu.com/news/sony-files-new-patent-that-mentions-accelerated-ray-tracing-on-ps5/
Patnet: https://data.epo.org/publication-server/document?iDocId=6759906&iFormat=0


5.3
Publication of the European patent application
5.3.001
The European patent application is published as soon as possible after the expiry of eighteen months from the date of filing or the earliest priority date. You may however request that it be published earlier.
The publication contains the description, the claims and any drawings, all as filed, plus the abstract. If the European search report is available in time, it is annexed (A1 publication); if not, it is published separately (A3 publication). If the European patent application was not filed in English, French or German, its translation will be published.
All European patent applications, European search reports and European patent specifications are published in electronic form only, on the EPO's publication server. The publication server is accessible via the EPO website (www.epo.org).




FMct7UJWQCE2mlk

FMct7TkWUAEYL8D

FMct7TkWUAEYL8D

FMct7T_WQBcPNlX

Ah, this will be the solution for Elden Ring framerate issues! Also it will spawn 455 videos of ray dancing comparisons.
 

Tripolygon

Banned
The patent does not talk only about an RT Unit like Intel or NV, it covers all bases of the implementation including using an RT unit for transversal and testing and using RT unit for transversal and shader for testing. An RT Unit in this regard is talking about ray transversal acceleration units in general.

In another aspect, a method for graphics
processing includesexecuting, ona graphicsprocessing
unit (GPU), a shader program that performs ray tracing
of a 3D environment represented by an acceleration
structure. The method also includes using a hardware-
implemented ray tracing unit (RTU) within the GPU that
traverses the acceleration structure at the request of the
shader program, and using, at the shader program, re-
sults of the acceleration structure traversal. The accel-
eration structure is a hierarchy with a plurality of levels
and acceleration structure traversal by the RTU includes
handling of coordinate transitions between the plurality
of levels of the acceleration structure.

[0018] In another aspect, a graphic processing unit
(GPU) includes at least one processor core adapted to
execute a software-implemented shader, and at least
one hardware-implemented ray tracing unit (RTU) sep-
arate from the processor core and adapted to traverse
an acceleration structure asynchronously with respect to
shader operation to identify intersections of rays with ob-
jects represented in the acceleration structure to gener-
ate results and return the results to the shader for iden-
tification by the shader of hits associated with the inter-
sections.

[0019] In another aspect, a graphic processing unit
(GPU) includes at least one processor core adapted to
executeasoftware-implementedshaderandatleastone
hardware-implemented ray tracing unit (RTU) separate
from the processor core and adapted to traverse an ac-
celeration structure to identify intersections of rays with
objects represented in the acceleration structure to gen-
erate results and return the results to the shader for iden-
tification by the shader of hits associated with the inter-
sections. The RTU includes hardware circuitry to imple-
menttraversallogictotraversetheaccelerationstructure
including modifying at least one ray to account for a
change in coordinate systems during traversal.
 
Last edited:

THE DUCK

voted poster of the decade by bots
Mark Cerny is a genius

Why do people keep saying this? He took basically of the shelf amd components, a fast ssd, and tweaked it a bit to keep the (ps5) machine cheap. We have seen almost no difference in performance between both next gen machines, aside from 1 game with fast loading levels that didn't really matter.
Whopdee doo.

That said, if he's come up with something that significantly reduces the overhead of rt, that might be something that affects performance in a larger way.
 
Last edited:

Snake29

RSI Employee of the Year
Why do people keep saying this? He took basically of the shelf amd components, a fast ssd, and tweaked it a bit to keep the (ps5) machine cheap. We have seen almost no difference in performance between both next gen machines, aside from 1 game with fast loading levels that didn't really matter.
Whopdee doo.

That said, if he's come up with something that significantly reduces the overhead of rt, that might be something that affects performance in a larger way.

Well that's exactly why he is genius...
 

Azelover

Titanic was called the Ship of Dreams, and it was. It really was.
We are way past good enough levels at this point.

This segment of the industry is ripe for disruption.
 

ChiefDada

Gold Member

Kinda confused about all of this. My initial thoughts were I don't see how it could be related to future hardware unless Sony is designing their own gpu or using a completely different approach to ray tracing. Otherwise, why would such details be in a Sony patent and not AMD?

On the other hand, the patent uses the term RT "unit" as opposed to RT "accelerator", the term AMD uses. Also says the rt units are separate from the gpu cores, which seems similar to nvidia approach.
 

Ezekiel_

Banned
Why do people keep saying this? He took basically of the shelf amd components, a fast ssd, and tweaked it a bit to keep the (ps5) machine cheap. We have seen almost no difference in performance between both next gen machines, aside from 1 game with fast loading levels that didn't really matter.
Whopdee doo.

That said, if he's come up with something that significantly reduces the overhead of rt, that might be something that affects performance in a larger way.
So you think building a console is simply like building a PC? You just select components and stick them in a box?

How clueless are you?

There's a great deal of customization in the PS5
 
On the other hand, the patent uses the term RT "unit" as opposed to RT "accelerator", the term AMD uses. Also says the rt units are separate from the gpu cores, which seems similar to nvidia approach.

no Sony nor AMD patent talks about "Ray Accelerators" concerning RDNA2. that's a term dubbed by marketing long after the design of AMDs RT hardware units was set in stone. you don't use marketing slang in patents.
 

Loxus

Member
Remember this is the same guy that was so adamant the PS5 didn't have hardware Ray Tracing.

avuqpPw.jpg

Also, there are multiple thing within a Compute Unit that are named with unit that doesn't have a separate dedicated block.
- Scalar Units
- Texture Filler Units
- Texture Mapping Units

In a patent you have to word your document in a way that you cover all basis from someone coping and wording it different.

If there is going to be a PS5 Pro, it would use the same Zen 2 CPU and RDNA 2 GPU (probably with an additional Shader Engine) similarly to how the PS4 Pro uses the same CPU and GPU with additional Compute Units.

Which means RT will be the same.
byrwW7c.jpg


Also, the PS6 isn't releasing until 2027 at least with these chip shortages.
The hardware from AMD that will be ready around that time probably isn't near being finalized and that's if development even started.

So how can this patent be not related to PS5?
 
Last edited:

Loxus

Member
Remember this is the same guy that was so adamant the PS5 didn't have hardware Ray Tracing.

avuqpPw.jpg

Also, there are multiple thing within a Compute Unit that are named with unit that doesn't have a separate dedicated block.
- Scalar Units
- Texture Filler Units
- Texture Mapping Units

In a patent you have to word your document in a way that you cover all basis from someone coping and wording it different.

If there is going to be a PS5 Pro, it would use the same Zen 2 CPU and RDNA 2 GPU (probably with an additional Shader Engine) similarly to how the PS4 Pro uses the same CPU and GPU with additional Compute Units.

Which means RT will be the same.
byrwW7c.jpg


Also, the PS6 isn't releasing until 2027 at least with these chip shortages.
The hardware from AMD that will be ready around that time probably isn't near being finalized and that's if development even started.

So how can this patent be not related to PS5?
To add to this,
Mark Cerny also calls it a unit in Road to PS5. Also, he calls it the Intersection Engine not Ray Accelerator.

"Another major new feature of our custom RDNA2 based GPU is Ray-Tracing.
f:id:keepitreal:20200329150003j:plain

Using the same strategy as AMD's upcoming PC GPUs.

The CUs contain a new specialized unit called the Intersection Engine, which can calculate the intersection of rays with boxes and triangles.

To use the Intersection Engine first you build what is called an acceleration structure.
f:id:keepitreal:20200329150107j:plain

It's data in RAM that contains all of your geometry.

There's a specific set of formats. You can use their variations on the same BVH concept. Then in your shader program, you use a new instruction that asks the intersection engine to check array against the BVH.

While the Intersection Engine is processing the requested ray triangle or ray box intersections, the shaders are free to do other work."


Edit:
Cerny using the word strategy, "Using the same strategy as AMD's upcoming PC GPUs."

Makes me think the Intersection Engine and Ray Accelerator might be two different RT implementations using a similar technique.
 
Last edited:

Fafalada

Fafracer forever
Interesting, was not sure what was in the RT units inside/alongside the TMU’s on RDNA2 (doing more in HW than just ray intersection tests).
The patent isn't describing RDNA2 - there's one figure showing that(simplified GPU Figure 3) - but the RTUs are a concept AMD hasn't introduced (yet).

No, this is for current hardware.
I'm really not seeing that? What I read seems pretty explicit about describing a hardware embodiment(a full hw implementation of acceleration structure traversal - I don't think even NVidia does that yet), not an algorithm that you'd implement in software. Traversal is something that happens in software on RDNA2, so there's only so much you can do async (execution units are a shared resource).
 
Tempest unit can assist in such calcs I presume

If it has the spare clocks to do so. Depends on the game logic which will change on a game-by-game basis.

Also Tempest is just a restructured single CU or dual-CU unit with certain graphics functions probably redundant for audio processes stripped out. So I don't think it will be that useful for assisting in what the patent is discussing.

I made you a GPU profile of WD:L running on a 6700XT to show you the problem with rdna2's RT (RT pipeline in red). i think its probably not "unaccelerated ray traversal" (which makes maybe 10% of the RT frametime) but that because of shared registries and logic interdependencies RT occupancy [in this case] is only at 25% for long portions of the frame.

wdlegion1440pultrahigxzjg0.png


i think that is the general problem this patent is trying to improve on.

So really it's more discussing a methodology to improve the process with technology already present in the hardware as-is, and not indicative of a new piece of hardware with architecture revisions to implement it?

I just scanned the patent so asking for clarification.
 

assurdum

Banned
Well There It Is Jurassic Park GIF







Our first hints at a PS5 Pro.
Jesus Christ why people give so much credit to Dictator who doesn't know a shit about how work intersection engine on ps5? This guy was blatantly ignorant about everything relative to the Sony hardware, don't believe to anything which coming to this mouth when a Sony console is involved. This guy admitted to not know a shit about GE on ps5 and meanwhile compared it to the GE on Vega GPU. He seems in first line just to downplay anything good relative to a PS5 news than really interested to discuss about it. Lately he is firmly invested to be a sort of tech guru on resetera but he is not always right with his straws and sometimes I read total absurdity in his posts.
 
Last edited:

ChiefDada

Gold Member
Jesus Christ why people give so much credit to Dictator who doesn't know a shit about how work intersection engine on ps5? This guy was blatantly ignorant about everything relative to the Sony hardware, don't believe to anything which coming to this mouth when a Sony console is involved. This guy admitted to not know a shit about GE on ps5 and meanwhile compared it to the GE on Vega GPU. He seems in first line just to downplay anything good relative to a PS5 news than really interested to discuss about it. Lately he is firmly invested to be a sort of tech guru on resetera but he is not always right with his straws and sometimes I read total absurdity in his posts.

Lol, in next DF Direct (Patreon) he says PS5 likely doesn't have a specific hardware feature that the Xbox Series X has and his only evidence is the fact that Sony hasn't mentioned it.

But when Sony mentions features they do have (ray tracing), he still doesn't believe it.


Confused Thinking GIF by MOODMAN
 

S0ULZB0URNE

Member
Why do people keep saying this? He took basically of the shelf amd components, a fast ssd, and tweaked it a bit to keep the (ps5) machine cheap. We have seen almost no difference in performance between both next gen machines, aside from 1 game with fast loading levels that didn't really matter.
Whopdee doo.

That said, if he's come up with something that significantly reduces the overhead of rt, that might be something that affects performance in a larger way.
rkNIHox.gif
 

Shmunter

Member
And I must comment here. You're not wrong, but touch on a very interesting point.

While that "sharing of solutions" may be a part of Sony's success. It's also the single largest contributor to Sony being in the corner they currently find themselves in, as well as why they're struggling to get out of it.

That corner I refer to is the one Sony has backed itself into. Where their 1st party studios are sharing so many solutions and tools that the only thing they do well is make singleplayer cinematic story driven games. That's not a bad corner to be in if that's the type of games you like to play. It certainly is though if you prefer other types of games.

It's been abundantly obvious for some time now that Sony's studios have been sharing a bit too much, for a bit too long.
Sony has neglected certain sectors of the gaming landscape no doubt. Their multiplayer & fps efforts have all but dried up. And they were big on these in the PS3 era. Now they need to buy Bungie to balance the deficit.

BUT, I’m commenting on technical matters previously, not gameplay design.
 

ckaneo

Member
What is he trying to say? AMD will be changing their raytracing unit to match intel/NV? Or Sony will move to intel or NV for the next PlayStation?
AMD will put out more ray tracing specific stuff in the future but it is unreleased, thus it's for future hardware
 
Last edited:

Shmunter

Member
If it has the spare clocks to do so. Depends on the game logic which will change on a game-by-game basis.

Also Tempest is just a restructured single CU or dual-CU unit with certain graphics functions probably redundant for audio processes stripped out. So I don't think it will be that useful for assisting in what the patent is discussing.



So really it's more discussing a methodology to improve the process with technology already present in the hardware as-is, and not indicative of a new piece of hardware with architecture revisions to implement it?

I just scanned the patent so asking for clarification.
I’m no Mark Cerney sipping champagne in a bath and merely putting 2 and 2 together on what’s been previously said.

The tempest cu unit has been repurposed as a stream/vector processor, very similar to ps3 spu. The result is a highly capable helper for crunching positional data on the fly, hence it’s ability to calculate 3D sound based on materials and position in a 3D environment.

All very reminiscent of casting rays around an environment to calculate light bounce. So to me at least, it makes theoretical sense.
 
Last edited:
And I must comment here. You're not wrong, but touch on a very interesting point.

While that "sharing of solutions" may be a part of Sony's success. It's also the single largest contributor to Sony being in the corner they currently find themselves in, as well as why they're struggling to get out of it.

That corner I refer to is the one Sony has backed itself into. Where their 1st party studios are sharing so many solutions and tools that the only thing they do well is make singleplayer cinematic story driven games. That's not a bad corner to be in if that's the type of games you like to play. It certainly is though if you prefer other types of games.

It's been abundantly obvious for some time now that Sony's studios have been sharing a bit too much, for a bit too long.

I mean what would you want the solution to be, then? Have the studios so at each other's throats leads from one threaten to leave if their tech is used by another? We don't need any further Yuji Naka incidents (he threatened to do this to Chris Penn and the Sonic Xtreme team; MASSIVE reason why Penn almost died from sickness from crunch and Xtreme getting cancelled).

We know part of Sony's roadmap includes live-service games, but they're in a unique position where they can bring all of the experience from their single-player projects into more multiplayer-centric projects. Just thinking of something like Forbidden West on a graphical/fidelity level (particularly the PS5 version) with the gameplay "loop" of a VALORANT or Apex Legends, or Destiny 2, would be genuinely massive and eye-catching.

And I think that's what they'll be doing, but with specific teams for those type of games i.e Bungie. But for that same reason I'm curious to see TLOU Part II: Factions because that will likely be a service game as well and Naughty Dog are arguably the best in the industry when it comes to character animations.
 

THE DUCK

voted poster of the decade by bots
Loading times and best looking games.

Loading time differences have been largely irrelevant in most games, so not really.
Best looking games so far this gen hasn't been a function of superior hardware, it's been a function of developers. We both know any of the ps5 games like demon souls, gt7 and Horizon dawn could all easily be done on series X without any issue whatsoever.
Even ratchet could be done if you take out the gimmicky instant level change that really wasn't needed - the game was great regardless.

Multiplatform games are basically a draw, how is that better?
 
Last edited:

Shmunter

Member
Loading time differences have been largely irrelevant in most games, so not really.
Best looking games so far this gen hasn't been a function of superior hardware, it's been a function of developers. We both know any of the ps5 games like demon souls, gt7 and Horizon dawn could all easily be done on series X without any issue whatsoever.
Even ratchet could be done if you take out the gimmicky instant level change that really wasn't needed - the game was great regardless.

Multiplatform games are basically a draw, how is that better?
Hardware is only part of a platform. The support and software created is the other. Hardware without the software is essentially landfill.
 

Beechos

Member
The chefs at both companys at it again i think last week there was a post about some new ms secret sauce for performance gains and this week its sonys turn.
 
I’m no Mark Cerney sipping champagne in a bath and merely putting 2 and 2 together on what’s been previously said.

The tempest cu unit has been repurposed as a stream/vector processor, very similar to ps3 spu. The result is a highly capable helper for crunching positional data on the fly, hence it’s ability to calculate 3D sound based on materials and position in a 3D environment.

All very reminiscent of casting rays around an environment to calculate light bounce. So to me at least, it makes theoretical sense.

But in practice Tempest isn't going to have the same processing load for non-audio RT tasks because at least some portion of it is going to be used for audio processing at any given moment. Maybe for audio that doesn't need the enhanced functions of 3D RT the system just leverages the general audio DSP; if that's the case then yeah theoretically it gives headroom for Tempest to be leveraged for graphical tasks like assisting in RT.

It's interesting in terms of theoretical uses; the only systems I know of where any audio component was used explicitly for anything outside of audio functions are the MegaDrive/Genesis (the Z80 used for audio could be used for Master System CPU with a Power Base Converter), and the Saturn (I think one of the Shining Force games had to use the audio processor for some game logic; not so sure if that included graphics-related functions).

It would be interesting to see a system architected with the explicit intent of leveraging advanced audio processing for non-audio graphics tasks actually be used with that scenario in practice, however. Maybe the PS5 is that type of system, we'll see.

Loading time differences have been largely irrelevant in most games, so not really.
Best looking games so far this gen hasn't been a function of superior hardware, it's been a function of developers. We both know any of the ps5 games like demon souls, gt7 and Horizon dawn could all easily be done on series X without any issue whatsoever.
Even ratchet could be done if you take out the gimmicky instant level change that really wasn't needed - the game was great regardless.

This is...partly true. Series X is in a weird predicament where apparently they're still having issues with some of the tools rollout. As that comes to past, performance across the board should generally improve.

However, and if I'm being 100% honest, there ARE changes some of those games you listed would need to make in order to run on Series X with the same framerate performance and general texture fidelity. PS5 does have a higher geometry culling and triangle rasterization rate, as well as higher pixel fillrate (helps with things like particle effects). And while it probably sounds tired out by now, having a pretty fast SSD that can stream in texture and asset data in a very short frame window does help with overall visual fidelity.

That's not to say Series X has a slow SSD; far from it. But between that, certain API tools still either being ironed out or not fully deployed, lack of certain design features that could've been useful (cache scrubbers for example) and certain overall design quirks (mainly the virtually split pools of RAM), it's not as easy as saying games optimized for a specific system like PS5 would just effortlessly translate over to a Series X with no adjustments made. And vice-versa for games specifically optimized for Series X's advantages (such as, potentially, mesh shader throughput) just put on PS5 with no adjustments made.



Multiplatform games are basically a draw, how is that better?
 
Last edited:

Shmunter

Member
But in practice Tempest isn't going to have the same processing load for non-audio RT tasks because at least some portion of it is going to be used for audio processing at any given moment. Maybe for audio that doesn't need the enhanced functions of 3D RT the system just leverages the general audio DSP; if that's the case then yeah theoretically it gives headroom for Tempest to be leveraged for graphical tasks like assisting in RT.

It's interesting in terms of theoretical uses; the only systems I know of where any audio component was used explicitly for anything outside of audio functions are the MegaDrive/Genesis (the Z80 used for audio could be used for Master System CPU with a Power Base Converter), and the Saturn (I think one of the Shining Force games had to use the audio processor for some game logic; not so sure if that included graphics-related functions).

It would be interesting to see a system architected with the explicit intent of leveraging advanced audio processing for non-audio graphics tasks actually be used with that scenario in practice, however. Maybe the PS5 is that type of system, we'll see.



This is...partly true. Series X is in a weird predicament where apparently they're still having issues with some of the tools rollout. As that comes to past, performance across the board should generally improve.

However, and if I'm being 100% honest, there ARE changes some of those games you listed would need to make in order to run on Series X with the same framerate performance and general texture fidelity. PS5 does have a higher geometry culling and triangle rasterization rate, as well as higher pixel fillrate (helps with things like particle effects). And while it probably sounds tired out by now, having a pretty fast SSD that can stream in texture and asset data in a very short frame window does help with overall visual fidelity.

That's not to say Series X has a slow SSD; far from it. But between that, certain API tools still either being ironed out or not fully deployed, lack of certain design features that could've been useful (cache scrubbers for example) and certain overall design quirks (mainly the virtually split pools of RAM), it's not as easy as saying games optimized for a specific system like PS5 would just effortlessly translate over to a Series X with no adjustments made. And vice-versa for games specifically optimized for Series X's advantages (such as, potentially, mesh shader throughput) just put on PS5 with no adjustments made.
Solid speculation, and all good point.

I’m a curious how much use tempest is actually getting right now outside of 1st party. Possible going cold most of the time.
 

THE DUCK

voted poster of the decade by bots
Hardware is only part of a platform. The support and software created is the other. Hardware without the software is essentially landfill.

Never stated that wasn't the case, what I stated was what cerny created wasn't anything substantially above the competition, certainly not "genius".
 
Last edited:

Zathalus

Member
The patent is partly discussing the acceleration of BVH traversal on a dedicated hardware unit, which the PS5 and all other RDNA 2 GPUs do not have ( including the Xbox Series). RDNA 2 does BVH calculations on the general compute units. This is the key reason why Ampere is so much faster with RT, as it does accelerate BVH calculations on the dedicated RT cores.

I'd speculate this is for the PS5 Pro or PS6.
 
So really it's more discussing a methodology to improve the process with technology already present in the hardware as-is, and not indicative of a new piece of hardware with architecture revisions to implement it?

I just scanned the patent so asking for clarification.

can't say for sure. could still be a method to tackle the general problems i describe only in PS5 Pro (which - if it exists - might be on rdna3 or 4) and not PS5.

but sure is that this patent claims a software side implementation and not hardware.
 
I’m no Mark Cerney sipping champagne in a bath and merely putting 2 and 2 together on what’s been previously said.

The tempest cu unit has been repurposed as a stream/vector processor, very similar to ps3 spu. The result is a highly capable helper for crunching positional data on the fly, hence it’s ability to calculate 3D sound based on materials and position in a 3D environment.

All very reminiscent of casting rays around an environment to calculate light bounce. So to me at least, it makes theoretical sense.
The problem is tempest has no direct access to CU caches (L0, L1 or even L2), as it can only access main ram (GDDR6).
 
RDNA 2 does BVH calculations on the general compute units. This is the key reason why Ampere is so much faster with RT, as it does accelerate BVH calculations on the dedicated RT cores.

this trope will never die...

if this was the case not just Ampere but also Turing would be monumentally faster in RT than rdna2 - which it is not.
the main reason why Ampere is so much faster in RT compared to rdna2 and Turing is, that it doubled the ray/triangle intersection test rate per clock.


it's really not good when vocal internet personalities just educate themselves about ray tracing based on nvidia marketing material.
 
Why do people keep saying this? He took basically of the shelf amd components, a fast ssd, and tweaked it a bit to keep the (ps5) machine cheap. We have seen almost no difference in performance between both next gen machines, aside from 1 game with fast loading levels that didn't really matter.
Whopdee doo.

That said, if he's come up with something that significantly reduces the overhead of rt, that might be something that affects performance in a larger way.

True that.

Xbox consoles are smaller, quieter, even have more features/performance (on paper as of now).

Yet nobody calls their designers genius, despite being better in every way, shape or form.

This is what marketing does to you, I guess.
 
Top Bottom