• 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.

Next-Gen PS5 & XSX |OT| Console tEch threaD

Status
Not open for further replies.

Imtjnotu

Member

looking season 7 GIF by Portlandia



TF am I looking for?
 
It wasn't the case because Sony is doing the same thing in their own way with Geometry Engine. Like was said before in this thread, RDNA 2 has no dedicated hardware for ML, which means MS can be using AMD's default implementation while Sony is once again using something of their own.
The PS5 GE isn't some secret sauce. AMD have had a GE since Vega. The PS5 uses Primitive Shaders, while the XSX and AMD RDNA 2 cards use Mesh Shaders.
Mesh Shaders are the next generation of shaders over Primitive.
The XSX has dedicated hardware to accelerate ML through lower precision calculations.
It wasn't stock RDNA 2.
At this point there is no evidence that Sony has INT 8 and INT 4 capability on their GPU.
It is what it is.
And even with the XSX having a more capable ML GPU, we haven't seen anything from MS about how much this is going to benefit the XSX.
So its not really something to put too much faith into yet.
 

kyliethicc

Member
I don't think MS could be listed as copying when they started work on it long before PS5 was out. The dates don't line up for it to be a copy. Was Sony copying Microsoft with rumble? Or achievements? lets be realistic.

Also the share button replaces how you used to access the share functions. again, not what I'd call a copy.
Copying Microsoft with rumble? lol what the fuck are you talking about

eH0Qnz0.jpg


Nintendo and Xbox both directly copied Sony's share button, and yes Sony directly copied Xbox achievements and called them trophies. Duh.
 
Last edited:
The PS5 GE isn't some secret sauce. AMD have had a GE since Vega. The PS5 uses Primitive Shaders, while the XSX and AMD RDNA 2 cards use Mesh Shaders.
Mesh Shaders are the next generation of shaders over Primitive.
The XSX has dedicated hardware to accelerate ML through lower precision calculations.
It wasn't stock RDNA 2.
At this point there is no evidence that Sony has INT 8 and INT 4 capability on their GPU.
It is what it is.
And even with the XSX having a more capable ML GPU, we haven't seen anything from MS about how much this is going to benefit the XSX.
So its not really something to put too much faith into yet.
A geometry engine is a technical term to describe the component of a GPU, PS5's "Geometry Engine" (uppercase G and E) is their own customisation they made to their GPU and is more custom then the standard RDNA 2 implementation and likely more advanced.

Both the Primitive Shaders and Mesh Shaders offer the same functionality and the same results, although they do have distinct differences in how they are implemented via software/API and Mesh Shaders are slightly more easier to code for when it comes to things like tessellation. Primitive and Mesh Shaders also use the same hardware and it's also the same across all RDNA 2 cards which also includes the PS5 and Series X/S. AMD are still filing patents for Primitive Shaders even this year.

It's true that Microsoft have put a greater emphasis on Machine Learning for things like backwards compatibility, however the hardware for ML that the Series X is using is also present on RDNA 1 (although disabled in early cards) and RDNA 2 and the only way for it not to be present on PS5 is if Sony intentionally removed or disabled it.
 

oldergamer

Member
A geometry engine is a technical term to describe the component of a GPU, PS5's "Geometry Engine" (uppercase G and E) is their own customisation they made to their GPU and is more custom then the standard RDNA 2 implementation and likely more advanced.

Both the Primitive Shaders and Mesh Shaders offer the same functionality and the same results, although they do have distinct differences in how they are implemented via software/API and Mesh Shaders are slightly more easier to code for when it comes to things like tessellation. Primitive and Mesh Shaders also use the same hardware and it's also the same across all RDNA 2 cards which also includes the PS5 and Series X/S. AMD are still filing patents for Primitive Shaders even this year.

It's true that Microsoft have put a greater emphasis on Machine Learning for things like backwards compatibility, however the hardware for ML that the Series X is using is also present on RDNA 1 (although disabled in early cards) and RDNA 2 and the only way for it not to be present on PS5 is if Sony intentionally removed or disabled it.
Actually if you look up amd vega geometry engine it will come up in a number of searches. What proof do we have that geometry engine is different from what was offered in vega?
 
Last edited:

kyliethicc

Member
I don't think he knows the difference between the DualShock and the PlayStation controller to be honest.
Well there were 3 main controllers for PS1, so I can see how someone could be confused. But he's not. He's just wrong.

Everyone knows Nintendo did the rumble pack and then Sony did the dualshock. Thats basic common knowledge. Sony didn't copy anything from xbox controllers, especially not rumble.
 
Last edited:
Well there were 3 main controllers for PS1, so I can see how someone could be confused. But he's not. He's just wrong.

Everyone knows Nintendo did the rumble pack and then Sony did the dualshock. Thats basic common knowledge. Sony didn't copy anything from xbox controllers, especially not rumble.

I remember owning an N64 with the rumble pack. You're right that it's absurd for anyone to say that Sony copied rumble from Microsoft. If anything they got the idea from Nintendo.
 
Actually if you look up amd vega geometry engine it will come up in a number of searches. What proof do we have that geometry engine is different what what was offered in vega?
The CUs in RDNA2 are also called CUs. What proof is there that the RDNA2 versions are different than those in Vega GPUs?

Do you see now obtuse this line of reasoning is?

(Perhaps even calling it "reasoning" is something of an overstatement)
 

Garani

Member
The CUs in RDNA2 are also called CUs. What proof is there that the RDNA2 versions are different than those in Vega GPUs?

Do you see now obtuse this line of reasoning is?

(Perhaps even calling it "reasoning" is something of an overstatement)
This is why the XSX is rumored to have CU from RDNA1. Which is pretty stupid, but a rumor is a rumor.
 
The CUs in RDNA2 are also called CUs. What proof is there that the RDNA2 versions are different than those in Vega GPUs?

Do you see now obtuse this line of reasoning is?

(Perhaps even calling it "reasoning" is something of an overstatement)

The abbreviation is CUs but both are different.

You see RDNA2 used Compute Units while RDNA1.5 uses Crap Units. The big difference between the two is that you need toilet paper to constantly clean out the CUs so they don't get dirty.

So oldergamer oldergamer is correct in that.

:messenger_winking_tongue:
 

HoofHearted

Member
Copying Microsoft with rumble? lol what the fuck are you talking about

eH0Qnz0.jpg


Nintendo and Xbox both directly copied Sony's share button, and yes Sony directly copied Xbox achievements and called them trophies. Duh.
Let’s give credit where credit is due.... ;)

”In 2002, Sony and Microsoft were sued by Immersion for patent infringement for the use of vibration functions in their gaming controllers.[1] Specifically, they were accused of infringing on claims in U.S. Patent 6,424,333 and U.S. Patent 6,275,213 (filed 2000 and 2001 as extensions of U.S. Patent 6,088,017, itself filed 1998, all "Tactile feedback man-machine interface device").[2] Both patents were continuation applications of a patent application originally filed in November 1995.[citation needed] Nintendo was not involved in the case, as the technology used in the Rumble Pak and GameCube controller (and, subsequently, the Wii Remote) is based on a different design, for which Nintendo holds the patents U.S. Patent 6,200,253 and U.S. Patent 6,676,520 based on a Japan patent application filed on October 9, 1995.”

 

kyliethicc

Member
Let’s give credit where credit is due.... ;)

”In 2002, Sony and Microsoft were sued by Immersion for patent infringement for the use of vibration functions in their gaming controllers.[1] Specifically, they were accused of infringing on claims in U.S. Patent 6,424,333 and U.S. Patent 6,275,213 (filed 2000 and 2001 as extensions of U.S. Patent 6,088,017, itself filed 1998, all "Tactile feedback man-machine interface device").[2] Both patents were continuation applications of a patent application originally filed in November 1995.[citation needed] Nintendo was not involved in the case, as the technology used in the Rumble Pak and GameCube controller (and, subsequently, the Wii Remote) is based on a different design, for which Nintendo holds the patents U.S. Patent 6,200,253 and U.S. Patent 6,676,520 based on a Japan patent application filed on October 9, 1995.”

who cares. oldergamer oldergamer is still completely fucking wrong.
 
Let’s give credit where credit is due.... ;)

”In 2002, Sony and Microsoft were sued by Immersion for patent infringement for the use of vibration functions in their gaming controllers.[1] Specifically, they were accused of infringing on claims in U.S. Patent 6,424,333 and U.S. Patent 6,275,213 (filed 2000 and 2001 as extensions of U.S. Patent 6,088,017, itself filed 1998, all "Tactile feedback man-machine interface device").[2] Both patents were continuation applications of a patent application originally filed in November 1995.[citation needed] Nintendo was not involved in the case, as the technology used in the Rumble Pak and GameCube controller (and, subsequently, the Wii Remote) is based on a different design, for which Nintendo holds the patents U.S. Patent 6,200,253 and U.S. Patent 6,676,520 based on a Japan patent application filed on October 9, 1995.”


What does that have to do with Sony copying rumble from Microsoft?
 

oldergamer

Member
The CUs in RDNA2 are also called CUs. What proof is there that the RDNA2 versions are different than those in Vega GPUs?

Do you see now obtuse this line of reasoning is?

(Perhaps even calling it "reasoning" is something of an overstatement)
Again do you have proof that its " custom". Im curious where that info some people are saying came from? outside of saying everything is "obtuse", got anything else to add?

We know xbox has fairly different cu's based on what has been made public.
 

assurdum

Banned
I am still not sure how RT works in RDNA gpus. Cerny had said that the RT hardware was built in the GPU. But apparently he wasn't talking about dedicated RT cores?

I am also not sure how Spiderman can do native 4k 30 fps with RT, better draw distance, more NPCs and yet has trouble doing that at half the resolution and twice the framerate. Is there a bottleneck somewhere? Why do they have to drastically reduce the NPC count and traffic density on top of halving the resolution of both the RT reflections and the base game? Whatever compute the shader processors were doing at native 4k should be the same amount of work at 1440p which is less than half the number of pixels in a native 4k image. And yet, despite the lower graphical settings, you have the resolution dipping all the way to 1080p.

If the bottleneck was on the CPU side with their variable clocks implementation, we wouldve seen it perform much worse in 120 fps modes in CoD and DMCV than the XSX. So it's not the CPU that is unable to do 60 fps with a far higher NPC count. So it must be their RT implementation that is creating a bottleneck where they are unable to run the game at the same graphical settings at half the resolution. Not a fan of playing spiderman in the 60 FPS performance RT modes. It just doesnt look as good.
What you don't understand? Raytracing is expensive as hell. Probably to have such raytracing at 60 FPS you have to go at 1080p with the amd card. Heck, even on Nvidia you need to stay many times around 1440p for 60 fps and it has even dedicated core which are an huge boost in raytracing performance.
 
Last edited:

assurdum

Banned
Again do you have proof that its " custom". Im curious where that info some people are saying came from? outside of saying everything is "obtuse", got anything else to add?

We know xbox has fairly different cu's based on what has been made public.
GE on ps5 from what reported different developers allows to have such access to the low API, to optimise easily tons of graphic stuff not possible before as ps5 is specifically designed around it. Now you can't believe is that big deal but many people won't believed ps5 was capable of raytracing in 4k and we know how is ended.
 
Last edited:

oldergamer

Member
GE on ps5 from what reported different developers allows to have access to the low API such to optimise easily tons of graphic stuff not possible before. Now you can't believe it but people won't believed ps5 was capable of raytracing in 4k and we know how is ended.
Not sure what u are getting at. I never said ps5 wasnt capable of something. Why cant i believeve it again?
 

oldergamer

Member
Uh you said GE on ps5 is like Vega GPU or something like that.
no i didnt say that. I said geometry engine was in vega and asked how do we know its different or that sony customized it in some way in ps5.

People keep talking about it like its a custom feature sony created. Im curious where that came from or if there is any public info that explains it.
 

assurdum

Banned
no i didnt say that. I said geometry engine was in vega and asked how do we know its different or that sony customized it in some way in ps5.

People keep talking about it like its a custom feature sony created. Im curious where that came from or if there is any public info that explains it.
No one has said Sony has invented the GE. The hell you are talking about. I explained to you as many others what have of so special. Handle the graphic in a game more easily. The prove you want it's in Spiderman Miles Morales. They have implemented raytracing in such short time (though is not exactly GE), it proves how easily can you manage complex graphic stuff on ps5 without years of development.
 
Last edited:
I feel like we've been over this whole GE thing for a while now, but I will quote an article by redgamingtech. This has been talked about to death.

"Inplace of Mesh Shaders, Sony is using the hardware-based Geometry Engine which essentially is Sony’s version of Mesh Shader, but totally accelerated via their own hardware block.

The GPU uses Primitive Shaders to address the GPU, along they’re apparently more complicated and more advanced than the ones found in RDNA 1. These primitive shaders can issue geometry instructions a little like Mesh Shaders and offer extreme precision on the GPU. This allows extreme culling of triangles super early on in the rendering pipeline of the GPU. The GE can be left to govern itself, but to really maximize the performance out of the console you’ll want to program it.

I was told now by multiple sources that the Playstation 5’s Geometry Engine is going to be even more critical for developing titles for the PSVR2, due the nature of pushing higher frame rates for VR solutions, and back when I was first leaking information on the PS5 development kit, this was also in line with those leaks too.

The Geometry Engine functionality is part and parcel of the custom elements of RDNA 2 from Sony, and VRS can run with extreme precision on the GPU too. We’ve already seen patents for Foveated Rendering (here is one) for the PS5 and which is used for the eye-tracking of PSVR2, and this seemingly leverages the GE of the console. It specifically mentions in the patents the varying resolution of the images based upon a users gaze.

The Playstation 5’s Geometry Engine though is a double-edged sword, as to my understanding squeezing the most out of the console will be smartly using the GE of the PS5. Again, the GE can govern itself, but you’re leaving a lot of optimization on the table, and developers aren’t even close to fully using the GE to its maximum potential.

Supposedly, the Geometry Engine has been so far best shown off in the Unreal Engine 5 Lumen demo, where it was highly leveraged along with “Nanite” to control geometry."
 

assurdum

Banned
I feel like we've been over this whole GE thing for a while now, but I will quote an article by redgamingtech. This has been talked about to death.

"Inplace of Mesh Shaders, Sony is using the hardware-based Geometry Engine which essentially is Sony’s version of Mesh Shader, but totally accelerated via their own hardware block.

The GPU uses Primitive Shaders to address the GPU, along they’re apparently more complicated and more advanced than the ones found in RDNA 1. These primitive shaders can issue geometry instructions a little like Mesh Shaders and offer extreme precision on the GPU. This allows extreme culling of triangles super early on in the rendering pipeline of the GPU. The GE can be left to govern itself, but to really maximize the performance out of the console you’ll want to program it.

I was told now by multiple sources that the Playstation 5’s Geometry Engine is going to be even more critical for developing titles for the PSVR2, due the nature of pushing higher frame rates for VR solutions, and back when I was first leaking information on the PS5 development kit, this was also in line with those leaks too.

The Geometry Engine functionality is part and parcel of the custom elements of RDNA 2 from Sony, and VRS can run with extreme precision on the GPU too. We’ve already seen patents for Foveated Rendering (here is one) for the PS5 and which is used for the eye-tracking of PSVR2, and this seemingly leverages the GE of the console. It specifically mentions in the patents the varying resolution of the images based upon a users gaze.

The Playstation 5’s Geometry Engine though is a double-edged sword, as to my understanding squeezing the most out of the console will be smartly using the GE of the PS5. Again, the GE can govern itself, but you’re leaving a lot of optimization on the table, and developers aren’t even close to fully using the GE to its maximum potential.

Supposedly, the Geometry Engine has been so far best shown off in the Unreal Engine 5 Lumen demo, where it was highly leveraged along with “Nanite” to control geometry."
A double edged sword for what reason? Optimization is fundamental especially in a close hardware. More possibilities of optimization you give to the developers, better the hardware will be pushed...
 
Last edited:
A double edged sword for what reason? Optimization is fundamental especially in a close hardware. More possibilities of optimization you give to the developers, better the hardware will be pushed...
He worded it poorly in the article but he did explain it better in his video, the GE although a powerful tool will take developers time to get grips with, he mentioned we won't see games fully utilising it till the second and third wave of games. Hardware engineers like LeviathanGamer2 also mentioned that things like the Primitive Shaders which Mark Cerny described sounded much more advanced than the RDNA 1 version and similarly RGT mentioned that the GE is like "Primitive Shaders on steroids" but is a little harder to get a handle on.
 
no i didnt say that. I said geometry engine was in vega and asked how do we know its different or that sony customized it in some way in ps5.

People keep talking about it like its a custom feature sony created. Im curious where that came from or if there is any public info that explains it.

I think in "Road to PS5" presentation with Cerny should shed a little more light on that question, if you're interested, below is the Video,




The word "Geometry Engine" has been around for quite some time, a small snippet from a Vega 10 white paper:

"Next-generation geometry engine To meet the needs of both professional graphics and gaming applications, the geometry engines in “Vega” have been tuned for higher polygon throughput by adding new fast paths through the hardware and by avoiding unnecessary processing. This next-generation geometry (NGG) path is much more flexible and programmable than before. To highlight one of the innovations in the new geometry engine, primitive shaders are a key element in its ability to..."​


Source: https://www.techpowerup.com/gpu-specs/docs/amd-vega-architecture.pdf


RDNA1 Whitepaper stuff:
(Referred to as Geometry Processor"

Also:
(Referred to as Geometry Engine/Geometry Processor)

"The 5700 integrates 40 RDNA compute units, a multi-level cache comprising an L2, L1, and L0, a geometry engine, 64 pixel units, and 4 asynchronous compute engines. One of the main design goals that AMD doubled down on was higher frequencies at lower power across all those components.​

Inside the graphics core of the die, in the center is the command processor which interfaces with the software over the PCI Express interface. Next to the command processor is the geometry processor which assembles the primitives and vertices. It is also responsible for distributing the work among the two shader engines. The two shader engines house all the programmable compute logic. Within each shader engine, there are two shader arrays. Inside each array is the primitive unit, a rasterizer, a group of four render backends (RBs), a shared L1 cache, and the new compute units."​



What does this prove? Nothing really, other than we can't say for certain that it is, or isn't cookie cutter AMD tech, I do find it odd, that Sony chose to highlight this specific feature in the developer talk, hell, there was some form of a geometry processor used in one or another form across every generation of PlayStation lol. The fact that they went out of their way to mention that, leads me to believe it's a custom solution, how custom? I don't know. Perhaps, it's off the shelf AMD tech that Sony feels it's team can take advantage/exploit in ways not done yet, or in ways that just makes optimization much easier.

I understand that people like to say "buuh buuh buug GE Engine, etc" when arguing system supremacy, and to a point I am usually eye rolling, but there is something happening under the hood, some secret sauce, that allows the PS5 to keep up with/exceed the XSX in games. Could it be this "GE" feature? Could it be cache scrubbers? Could it be the faster/narrower GPU? Could it be the alleged unified cache for the Zen2 CCX's? Who knows. It's anyone's guess, and those facts/rumors can be a rabbit hole of arguments, but results are really hard to argue with, and the results, for the PS5 as is, is rather surprising.
 
I think in "Road to PS5" presentation with Cerny should shed a little more light on that question, if you're interested, below is the Video,




The word "Geometry Engine" has been around for quite some time, a small snippet from a Vega 10 white paper:

"Next-generation geometry engine To meet the needs of both professional graphics and gaming applications, the geometry engines in “Vega” have been tuned for higher polygon throughput by adding new fast paths through the hardware and by avoiding unnecessary processing. This next-generation geometry (NGG) path is much more flexible and programmable than before. To highlight one of the innovations in the new geometry engine, primitive shaders are a key element in its ability to..."​


Source: https://www.techpowerup.com/gpu-specs/docs/amd-vega-architecture.pdf


RDNA1 Whitepaper stuff:
(Referred to as Geometry Processor"

Also:
(Referred to as Geometry Engine/Geometry Processor)

"The 5700 integrates 40 RDNA compute units, a multi-level cache comprising an L2, L1, and L0, a geometry engine, 64 pixel units, and 4 asynchronous compute engines. One of the main design goals that AMD doubled down on was higher frequencies at lower power across all those components.​

Inside the graphics core of the die, in the center is the command processor which interfaces with the software over the PCI Express interface. Next to the command processor is the geometry processor which assembles the primitives and vertices. It is also responsible for distributing the work among the two shader engines. The two shader engines house all the programmable compute logic. Within each shader engine, there are two shader arrays. Inside each array is the primitive unit, a rasterizer, a group of four render backends (RBs), a shared L1 cache, and the new compute units."​



What does this prove? Nothing really, other than we can't say for certain that it is, or isn't cookie cutter AMD tech, I do find it odd, that Sony chose to highlight this specific feature in the developer talk, hell, there was some form of a geometry processor used in one or another form across every generation of PlayStation lol. The fact that they went out of their way to mention that, leads me to believe it's a custom solution, how custom? I don't know. Perhaps, it's off the shelf AMD tech that Sony feels it's team can take advantage/exploit in ways not done yet, or in ways that just makes optimization much easier.

I understand that people like to say "buuh buuh buug GE Engine, etc" when arguing system supremacy, and to a point I am usually eye rolling, but there is something happening under the hood, some secret sauce, that allows the PS5 to keep up with/exceed the XSX in games. Could it be this "GE" feature? Could it be cache scrubbers? Could it be the faster/narrower GPU? Could it be the alleged unified cache for the Zen2 CCX's? Who knows. It's anyone's guess, and those facts/rumors can be a rabbit hole of arguments, but results are really hard to argue with, and the results, for the PS5 as is, is rather surprising.


Seems like programmability is a key feature to culling with the GE. I remember Cerny clearly stating that it was fully programmable in the PS5. Maybe this is what the customizations are with the GE?

PlayStation 5 has a new unit called the Geometry Engine which brings handling of triangles and other primitives under full programmatic control.
 
Again do you have proof that its " custom". Im curious where that info some people are saying came from? outside of saying everything is "obtuse", got anything else to add?

We know xbox has fairly different cu's based on what has been made public.

It’s clear by the wording and mention by Mark Cerny in his Road to PS5 talk. It’s also been corroborated by media outlets like RGT who are hearing the same things from devs.

The point of my previous post is that even AMD uses the same wording for IP blocks of their GPUs between micro-architectural generations. Just because the nomenclature is the same doesn’t mean the IP blocks themselves haven’t changed or been updated between generations. In fact, in most cases, when there is a big microarchitectural jump, like from GCN to RDNA, almost everything is different, and is so by definition; as how do you expect a far more capable Graphics and Compute Array in RDNA to be fed by a Geometry Engine designed for a much weaker GPU microarchitectural design, I.e. Vega. It’s just common sense.

The PS5 Geometry Engine is clearly being put forward in Cerny’s talk as a new innovation distinct from standard desktop RDNA iterations. How do we know? Because everything else he mentions in the same talk is also; I.e. the Tempest Engine, their bespoke I/O system, GPU cache scrubbers etc.

On the balance of probabilities, there’s little reason to believe it isn’t distinct in some way. What way exactly?

Programmability... see MasterCornholio MasterCornholio ’s post below.

Some of u guys are overly defensive about everything

Nah, we’re just allergic to the smell of BS, misinformation and the FUD posters like you make a career out of.

Seems like programmability is a key feature to culling with the GE. I remember Cerny clearly stating that it was fully programmable in the PS5. Maybe this is what the customizations are with the GE?

This. Every previous AMD and NVidia GPU has a fixed function geometry processing unit that is responsible for triangle setup from the vertex data produced by the CPU. The Geometry Engine in Vega and RDNA encompasses this geometry processor as well as scheduling logic for scheduling work items to the shader array.

In Vega, the GE included the facility for devs to write Primitive Shaders, which were supposed to allow for some limited processing of geometry earlier in the rendering pipeline. An example of this would be primitive back-face culling, that Vega’s Primitive Shaders were supposed to be many times faster at than the fixed function culling hardware on previous GCN GPUs. Unfortunately, in Vega, the PS were disabled and the fallback path using the existing fixed function hardware for culling of tris was so bad (actually worse than Polaris) that devs mostly did their culling using compute shaders (meaning culling was being done later in the pipeline and so unnecessary processing would be done on tris that would later be culled, thus losing efficiency).

From the sounds of Cerny’s comments as well as the reports from devs and media outlets. Not only does the PS5 GE include the possibility to employ Primitive Shaders, but this capability is brought into dev’s full programmatic control. In addition, Cerny’s comments hint towards additional functionality encompassing the function of Mesh Shaders on NVidia hardware.

Personally speaking, and you can feel free to take this with a grain of salt as you wish, i’ve heard dev comments that allude to the fact that the PS5 GE can allow for geometry to be generated on the fly (rather than just being derived from data generated by the CPU). So in theory, the PS5 can process geometry in a way that isn’t limited by the capability of the fixed function geometry setup hardware in the GPU front end, nor the memory bandwidth limitations of what vertex data you can pull in from main memory. Tessellation is a form of this, but Tessellation relies on subdivision of an existing mesh, whereas the PS5 GE is able to create whole new geometry from Shaders written by the developer. If you listen to Mary Cerny’s wording in his PS5 presentation, he basically confirms this.

In Cerny’s talk he lists possible uses of the GE:

Simple:
- Culling of primitives

More advanced:
- Synthesising geometry on the fly

Cerny cites, the latter can be applied to for e.g. smoothly varying LOD transitions (way more sophisticated than Tessellation alone can achieve), procedural generation of fine detail on close up objects, and improvements to particle effects and other visual special effects.

All three example applications fall well beyond the scope of what Vega’s Primitive Shaders can do.
 
Last edited:

Hendrick's

If only my penis was as big as my GamerScore!
Inevitable chasing lol. Xbox copied Sony over last 2 gens by eventually adding the headset jack, Bluetooth, (optional) rechargable batteries and a USB port. Now they're just getting around to copying the share button. Xbox will add haptics and resistant triggers to their controller over the generation, maybe they'll even finally add a gyro and or a touchpad.
That patent was from 2018. Try harder.
 

3liteDragon

Member
The PS5 GE isn't some secret sauce. AMD have had a GE since Vega. The PS5 uses Primitive Shaders, while the XSX and AMD RDNA 2 cards use Mesh Shaders.
no i didnt say that. I said geometry engine was in vega and asked how do we know its different or that sony customized it in some way in ps5.

People keep talking about it like its a custom feature sony created. Im curious where that came from or if there is any public info that explains it.
No one said it was a secret sauce feature, what's special about PS5's custom GE (according to RGT and MLiD) is the fact that it can cull geometry way earlier in the pipeline compared to PC RDNA 1 and 2, and will be adopted by AMD for PC RDNA 3. And according to MLiD, the way PS5's GE handles VRS is better than how PC RDNA 2 and the Series consoles handle it, which is why the PC implementation of VRS isn't in PS5.

All the videos below are timestamped.





 
Last edited:
Status
Not open for further replies.
Top Bottom