• 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.

Digital Foundry about XSX teraflops advantage : It's kinda all blowing up in the face of Xbox Series X

PaintTinJr

Member
The concept of hardware acceleration is usually referred to a unit designed specifically to run certain instructions. Shaders are generalist parallels processors, so they are not hardware accelerators.
For example, if a GPU has a unit specific to decode AV1, that is hardware acceleration decode/encode of AV1 video. But if it's run on shaders, the it's just software based.
An example of this is relating to UE5 is rasterization. A GPU has specific units for rasterization, what you called ASICs. But EPIC choose to do software rasterization, because it is more flexible and better suited for their engine.

In the the case of RT, nvidia has units to specifically accelerate both, BVH traversal and ray testing.
In the case of RDNA2, it only has some instructions inside the TMUs, that accelerate ray-testing. But BVH traversal is done in software, in the GPU shaders.
RDNA2 does not have any hardware to accelerate BVH traversal. It's just shaders and they don't even have instructions to accelerate the BVH.



I should have been clearer about what I was talking about when refereeing to Mesh Shaders.
I'm talking about the new GPU pipeline for geometry rendering, introduced with DX12_2
Previously in DX12, we had these stages: Input Assembler; Vertex Shader; Hull Shader; Tessellation; Domain Shader; Geometry Shader; Rasterization; Pixel Shader
But with the new pipeline: Amplification Shader; Mesh Shader; Rasterization and Pixel Shader.
It's a simpler pipeline that reduces overhead and increases geometry throughput significantly.

GPUs have been doing hardware culling, to prevent overdraw, even before the existence of programable shaders.
Of course it wasn't as advanced as what we have today, but it did offer performance improvements.



We already talked about this. That feature is called In-Line ray-tracing.
Something that the Series X and RDNA2 on PC can do. Even NVidia's hardware benefited from this, as it improved contention during the execution pipeline.

BTW, can you point me to where Cerny said that.



But geometry can be culled using an Hierarchical Z:
That is what the r.HZBOcclusion cvar does in Unreal.

This is not related to UE5, but it's a good example of occlusion culling with Hierarchical-Z
In that Hierarchical-Z article it is all in frustum culling, as I previously said, with the part below about a pre-pass stating as such

Frustum Culling Pre-pass​

Before attempting any kind of Hi-Z culling, we first attempt to frustum cull the instance. Frustum culling is implemented very efficiently (especially for bounding spheres), and by frustum culling, we also avoid some edge cases with the Hi-Z algorithm later.

This algorithm will provide additional saving in the nanite pass - which is done in a compute shader according to Epic using mesh shaders - but in relation to my point about real-time kit-bashing, this algorithm will be doing the self occluding saving from the part of the kit-bashing where part of the set theory (difference) resulting BVH contains part of one megscan's geometry inside the result..... they could remove that using a succession of kit-bashes to leave just the outer shell of the union: (edit see blender mock up showing the difference between A+B, versus A not B, B not A, A not B + B not A. from left to right

K6VGRhy.jpg





But using HiZ would probably be quicker than 2 extra kit-bashes per mesh, when occlusion culling it is getting done anyway.

As for suggesting the RT on AMD isn't hardware accelerated, your problem is that you are incorrectly attributing Nvidia with having provided units that definitively solve the problem - in the way a AV transcoder does for a codec specification - when research into real-time RT is still in its infancy and the AMD solution will long out live the RTX unit and DX RT restricted solutions in versatility. Yes, the RTX units provide great performance in current games using RT because of the separate unit nature, silicon area afforded to them and the ability to execute in fewer clock cycles for the same quantity of rays, but that in no way alters that the AMD hardware does provide hardware accelerated ray tracing even if it occupies much of the task in a shader.

As for the Cerny quote from Road to PS5 regarding RT hardware this is from the transcript.
Its 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.

As for the new DX pipeline, it is still the same hardware underneath with the vendor(Nvidia) GPU assembly just providing that abstraction. it will only be fixing up shortcomings in DX that aren't in the PS5 GPU access so the point I was making still remains unchanged about the geometry pipeline.
 
Last edited:

SHA

Member
Right, what matters games that looks and plays different, rehashing kills the basic fundamental of Technology, games should look different from the first place.
 

onQ123

Member


In text
It's probably easiest to talk about Tempest-based 3D Audio and the Dolby device support in terms of Ambisonic audio, which is increasingly popular these days (note there are other strategies for 3D Audio, including ones that use discrete 3D audio objects, but situation is rather similar).

Ambisonic audio can be viewed as a pretty radical extension of stereo audio. With stereo audio, the game's audio engine (or the middleware being used) will add a sound source into one or both channels based on its location - if the source is to the right of the listener it's primarily added into the right channel, and so on. With Ambisonic audio, there are a lot more channels - fifth order is very common and uses 36 channels, so it allows pretty good locality to the audio.

A sound source is then added into those 36 channels based on location; the math is a bit more complex than when using stereo but not overwhelmingly so. Because the audio processing is channel based (albeit at 36 channels rather than 2 channels), the audio designer keeps very good control of mixing, filters, etc., and strategies like dynamic range compression (where audibility of certain important audio such as player character voice is ensured] can be used as usual.

The Ambisonic audio channels are then handed off to the Tempest 3D Audio Tech engine for rendering, which is to say that the Tempest engine uses the player's HRTF and the speaker locations to create an appropriate audio stream for each speaker. The Ambisonic audio channels encode all directions, including above the player; even if rendering for headphones, this is very important, because it allows a sound "above" the player to be processed in such a way to sound as if it is truly coming from above - this is of course where the HRTF with its encoding of head and ear shape comes in.

Up until the most recent update, the Tempest engine would render the information in the Ambisonic channels into headphones, stereo TV speakers, and 5.1 and 7.1 audio setups. Now 7.1.4 has been introduced, with its four overhead speakers, but really nothing changes in the overall Tempest rendering strategy - the 36 Ambisonic channels already include audio coming from all directions, including above the player. To put that differently, the support of the four overhead speakers is "first class" support, they are

Also note the rendering latency for these new speaker setups is identical to what it has been in the past for stereo, 5.1 and 7.1. As a result, the 7.1.4 experience for existing games should be quite good. It is true that the game teams could not test with these speaker setups but support should be pretty automatic, the necessary game audio data is already there in Ambisonic form. Going forward, there's an opportunity for improvement as the sound designers can verify the highest quality of audio on 7.1.4 speaker setups as well.

- Mark Cerny, Lead System Architect of the PS5
 
Last edited:

rofif

Can’t Git Gud
Let's not forget that Asus ROG handheld boasts 8,4tf power too...
And as far as I am know, it is not a ps5/2070super device. Not even connected. It barely does 720p30 in modern games
 

winjer

Gold Member
Let's not forget that Asus ROG handheld boasts 8,4tf power too...
And as far as I am know, it is not a ps5/2070super device. Not even connected. It barely does 720p30 in modern games

Those 8.4 TFLOPs are marketing numbers. Reality is very different.

Two things to consider. Asus advertises a GPU clock of up to 2.4Ghz. In reality, the GPU clocks is closer to 1.7Ghz.
The other thing to consider is that no game uses the double issue units of RDNA3. So half of the shaders in it are not used.
So with single issue and only 1.7Ghz, the result is 2.45 TFLOPs.

Meanwhile the 10 TFLOPs for a PS5 are correct.
 

sinnergy

Member
Probably important enough to deserve its own thread. Just due to all the discussions we had on it.
Richard basically says game engines need to rebuild or new engines , he is correct, but this is also true for Series for these features mesh shaders /SFS. My opinion still stands , all engines are old and we don’t see big improvements yet, maybe in 2 years or so. My opinion is also : PS5 benefits more from the engines now used because of the faster CPU clock.
 
Richard basically says game engines need to rebuild or new engines , he is correct, but this is also true for Series for these features mesh shaders /SFS. My opinion still stands , all engines are old and we don’t see big improvements yet, maybe in 2 years or so. My opinion is also : PS5 benefits more from the engines now used because of the faster CPU clock.

I don't believe in any magical power boost. The two will remain with similar results that they currently have. It will be up to next gen consoles to show a big difference if there are any.

P.S Excluding any mid gen consoles of course.
 

Darius87

Member
I don't believe in any magical power boost. The two will remain with similar results that they currently have. It will be up to next gen consoles to show a big difference if there are any.

P.S Excluding any mid gen consoles of course.
this has nothing to do with power think more like psvr foveated rendering for flatscreens.
 

John Wick

Member
It’s always been a strategy problem with Microsoft.

The hardware features that make Series consoles special simply aren’t being utilized. If every game was designed to make use of SFS, VRS2, DirectML, Mesh Shaders etc. the power disparity would be obvious.

The SDK itself is intended on helping developers release on both Xbox / PC. There is no incentive whatsoever to integrate the most advanced Series only features. That’s on Microsoft and Phil Spencer’s crusade of developer appeasement. Even 1st party devs don’t use these features.

Microsoft should have just built a simpler box designed to brute force everything. The Series X hardware is almost wasted as everything special about it, that gives it an identity is being ignored.
The hardware features that make SSX so special aren't being used???
The so called full spec RDNA2 features are built into every AMD GPU. Nvidia has it's own versions named differently. They are being used but aren't the game changers you thought they would be. PS5 has it's own special hardware features and it's getting better as well just as Series consoles improve.
 

Alebrije

Member
Since PS4 Sony has a better team team to develop a console vs Microsoft.

Besides power you need performance and PS5 behaves more like a solid unit where all is calculated to avoid bottleneks.
 
Last edited:

Dorfdad

Gold Member
Richard basically says game engines need to rebuild or new engines , he is correct, but this is also true for Series for these features mesh shaders /SFS. My opinion still stands , all engines are old and we don’t see big improvements yet, maybe in 2 years or so. My opinion is also : PS5 benefits more from the engines now used because of the faster CPU clock.

This is why we will see more and more developers, moving towards unreal engine five these things take years and years to develop and most studios. Don’t have the time nor resources to continue to develop an update these engines to take full advantage of updated systems.
 

Killjoy-NL

Member
I can't fault them for believing the Series X would have slightly higher performance than the PS5. The specs on paper would reasonably lead most to that same conclusion.


But you can't argue with the results we've seen 3 years in. PS5 is a better performing piece of hardware more often than not.
How so?
Cerny was pretty clear in Road to PS5 in a way that even people with limited understanding of hardware could understand that there would be similar performance.
 
How so?
Cerny was pretty clear in Road to PS5 in a way that even people with limited understanding of hardware could understand that there would be similar performance.

Sure, but what makes more sense for a gaming hardware analysis channel to focus on: The hard specs that are presented for each platform or the words of someone that works for Sony?

Again, clearly the results have quantified Cerny's claims but you can't blame a technical analysis group for focusing on technical specifications.
 

Killjoy-NL

Member
Sure, but what makes more sense for a gaming hardware analysis channel to focus on: The hard specs that are presented for each platform or the words of someone that works for Sony?

Again, clearly the results have quantified Cerny's claims but you can't blame a technical analysis group for focusing on technical specifications.
Since the hard specs turned out to be wrong and, going by the leaked MS emails, even Xbox agreed with the words of someone that works for Sony, it's at least utter incompetence from DF.

Again, Cerny explained it in a very easy to understand way and even developers praised PS5 architecture, while the Xbox-camp was cheering about the Tflop count.
 
Last edited:
DF flaming the console war



I would also be asking Microsoft why they haven’t done more to showcase their consoles capabilities compared to PlayStation with their own first party offerings because in my opinion they haven’t done enough.

You couldn’t shut them up about power up to Series X’s launch and now they barely mention it, it’s marketing focuses far more on Series S.

I hope more consumers hold off on a day one/ launch window hardware release purchase from them in future.
 
Last edited:

PaNaMa

Banned
Honestly those 4 fps PS5 usually hold over Series comes down to clock speed - PS5 has a higher boost clock and is basically always boosting. MS has locked lower clocks. with extra CUs. That was the tradeoff they chose, but the way devs program things clock speed seems to always matter more. Someone smarter than me can speak to it better I'm sure. I'd love to see MS do a lab / beta firmware with a 200Mhz clock speed bump and see if the gap goes to Zero.
 
Since the hard specs turned out to be wrong and, going by the leaked MS emails, even Xbox agreed with the words of someone that works for Sony, it's at least utter incompetence from DF.

Again, Cerny explained it in a very easy to understand way and even developers praised PS5 architecture, while the Xbox-camp was cheering about the Tflop count.

What? The leaked e-mails showed that Phil was confident that Xbox was the better piece of hardware compared to PS5:

Unredacted reaction to PS5's reveal, "We have a better product than Sony does" - Phil Spencer 30% GPU advantage
 

DeepEnigma

Gold Member
DF flaming the console war


Xbox has a forced parity. It was revealed in the Spencer emails.

It's probably why the RDR remaster doesn't have unlocked framerate on the PS5 or PS4Pro, since R* would have to be forced to patch the BC version on Xbox. And they are not about to open back up that spaghetti western code on the original game.
 
Last edited:
Top Bottom