Support NeoGAF

[Topher]

Where is @Riky these days?

Was banned a few weeks ago and never returned.

No account suicide thread or nothing.

With every post of yours, I'm more and more certain you are some sort of bot. I have so many questions, but sadly you can't answer them.

At least he isn't replicating.

 

[El Pistolero]

PS5 is a great console.

Almost as if Cerny knows what he is doing!

 

[ChiefDada]

The presentation doesn't say either is better or worse, just that they had to optimize differently for Xbox compared to PS5 because of difference in architecture and pipelines.

Huh? You literally posted quotes confirming everything that I said. PS5 RT was faster/more efficient than Series consoles.

One solution to that problem that we have for techniques that use ray tracing shaders on XSX is to use what we call “Specialized State Objects”:
Which means the high-level ray tracing pipeline is broken down internally into multiple different ray tracing pipelines based on ray-gen VGPR
And then we select the one we want to avoid paying the occupancy penalty all the time.
And this is not needed on PS5 because there are no ray tracing pipelines as such.
Each ray gen shader is just a regular compute shader with known resource allocations.

The "Specialized State Object" for Series optimization is to reduce this inefficiency. It clearly didn't exceed PS5 because otherwise they would have optimized for PS5 as well. Do you need them to explicitly state that "PS5 was still faster even after all of these optimizations"? They are a 3rd party engine developer; that will never, and should never, happen.

 

[Tripolygon]

Huh? You literally posted quotes confirming everything that I said. PS5 RT was faster/more efficient than Series consoles.



The "Specialized State Object" for Series optimization is to reduce this inefficiency. It clearly didn't exceed PS5 because otherwise they would have optimized for PS5 as well. Do you need them to explicitly state that "PS5 was still faster even after all of these optimizations"? They are a 3rd party engine developer; that will never, and should never, happen.

Could also just be a case of DXR being inefficient and not specifically a hardware issue but that is way beyond my understanding.

 

[adamsapple]

Huh? You literally posted quotes confirming everything that I said. PS5 RT was faster/more efficient than Series consoles.

And in the later slides they mentioned how they overcame the difference in architecture and pipelines via optimization. That's the whole point.

For someone who said you don't want to console war, you reached that conclusion very quick

It works differently for different games/engines/optimization levels, for example Control's RT mode has a 16% performance advantage on SX versus PS5.

 

[Ghost of Caelid]

Wait til you find out about the PS5 Digital Edition!

I have that one...its a great console too.

 

[ChiefDada]

I wonder if that is a hardware issue or an inefficiency with the DXR API.

Could also just be a case of DXR being inefficient and not specifically a hardware issue but that is way beyond my understanding.

YES!! That is precisely my point from the start. API plays a huge role. I personally believe hardware isn't the main reason for performance differences going forward.

API also plays a role. In the recent UE5 presentation, we learned that PS5 ray tracing was faster/better suited for Lumen Hardware RT than Series X/S. Not because of any issue with hardware, but because DXR was inefficient and PS5's was more flexible.

 

[ChiefDada]

And in the later slides they mentioned how they overcame the difference in architecture and pipelines via optimization. That's the whole point.

For someone who said you don't want to console war, you reached that conclusion very quick

It works differently for different games/engines/optimization levels, for example Control's RT mode has a 16% performance advantage on SX versus PS5.

They didn't "overcome" it, but they did have a solution to minimize it. They still have to deal with these "occupancy penalties" occuring under XS RT pipeline, just much less of it; they decide which ones they are willing to incur. Again, you quoted it yourself:

... And then we select the one we want to avoid paying the occupancy penalty all the time.

But whatever dude, I give up. You see what you want to see.

 

[adamsapple]

They didn't "overcome" it, but they did have a solution to minimize it. They still have to deal with these "occupancy penalties" occuring under XS RT pipeline, just much less of it; they decide which ones they are willing to incur. Again, you quoted it yourself:



But whatever dude, I give up. You see what you want to see.

Keep in mind that presentation is based on the one isolated case of The Matrix Awakens. We need more UE5 based games with RT, Lumen, Nanite on both consoles to get a full picture of how developers are tackling it.

 

[Tripolygon]

YES!! That is precisely my point from the start. API plays a huge role. I personally believe hardware isn't the main reason for performance differences going forward.

Thanks for sharing the Presentation, i read these siggraph presentations, it is a treasure trove on at least getting familiar with game development.

 

[Putonahappyface]

Was banned a few weeks ago and never returned.

No account suicide thread or nothing.

At least he isn't replicating.

 

[Dafegamer]

Inb4 "where the 12TF at"

 

[Darsxx82]

It's the engine, all the recent CoD games in the last few years have had the same result. Identical in 60hz and runs a little better on PS5 in the 120hz mode.

XBO X is also downright poorly optimized on all these recent COD games (on SP and MP modes). It shows quite disappointing performance against PS4Pro....

I think it's more optimization than an engine thing.

 

[Mr Moose]

XBO X is also downright poorly optimized on all these recent COD games (on SP and MP modes). It shows quite disappointing performance against PS4Pro....

I think it's more optimization than an engine thing.

It's because the resolution is too high.

PS4 Pro uses a dynamic resolution with the highest resolution found being approximately 2730x1536 and the lowest resolution found being approximately 1365x1536. PS4 Pro uses a form of temporal upsampling that can reconstruct an approximately 2730x1536 resolution when rendering below this resolution. PS4 Pro renders the UI at 3840x2160.

Xbox One X uses a dynamic resolution with the highest resolution found being 3840x2160 and the lowest resolution found being 1920x2160. Xbox One X renders at 3840x2160 much less often than PS5 and Xbox Series X do in their 60fps mode. Xbox One X uses a form of temporal upsampling that can reconstruct a 3840x2160 resolution when rendering below this resolution.

 

[Lysandros]

Not surprised honestly. I've been saying it since before launch, it's gonna go back-and-forth.

As to why... honestly, its probably as "simple" as some engines just work better with one machine over the other. This time PS5's smaller faster pipes work better than the XSX's fatter slower pipes, I make this assumption based on all the tech talk I've seen about the machines on here.

There are actually 'pipes' which are both 'fatter' and faster, like ROPs of PS5.

 

[Darsxx82]

It's because the resolution is too high.

XBO X is much more powerful than PS4Pro. If they lower the resolution you would get an equality that does not exist at the power level.

A good optimization on XBO X should result in equal performance and more resolution, or both. In the last COD games we have never seen a good optimization on XBO X.

 

[itsjustJEFF]

There are actually 'pipes' which are both 'fatter' and faster, like ROPs of PS5.

Yea, got that term for the tech guys in here lol

 

[Mr Moose]

XBO X is much more powerful than PS4Pro. If they lower the resolution you would get an equality that does not exist at the power level.

A good optimization on XBO X should result in equal performance and more resolution, or both. In the last COD games we have never seen a good optimization on XBO X.

It's not twice as powerful though, check the math.
1365x1536-2730x1536
1920x2160-3840x2160
They are simply aiming too high.

 

[Lysandros]

Honestly the difference in 120 FPS mode appears to be larger than i thought, PS5 is up to 10 FPS ahead on at points on top of running with full Vsync compared to XSX. There some are reports of TAA having higher precision on PS5 with less ghosting based on the analysis, to be confirmed. Series S uses an even lower implementation apparently it is obvious in the video, i didn't hear Tom mentioning it.

 

[Darsxx82]

It's not twice as powerful though, check the math.
1365x1536-2730x1536
1920x2160-3840x2160
They are simply aiming too high.

There are many games that have shown a 2x advantage in favor of XBO X.

But it doesn't matter, the fact that XBO X doesn't have the same or better fps performance than PSPro is an example of bad optimization. A good optimization on XBO X should achieve the same or better fps than on PSPro and surely it would still have room to present a higher resolution.

 

[TheRedRiders]

Keep in mind that presentation is based on the one isolated case of The Matrix Awakens. We need more UE5 based games with RT, Lumen, Nanite on both consoles to get a full picture of how developers are tackling it.

I would be curious to see this as well.

I would have thought UE5 would perform better on Series X compared to the PS5 because the engine is compute heavy (which favours the Series X). Yet the results we got for the Matrix demo show almost equal parity between the two systems.

 

[nowhat]

This time PS5's smaller faster pipes work better than the XSX's fatter slower pipes

So in some instances, a small and fast pipe is superior. I have been validated.

 

[adamsapple]

So in some instances, a small and fast pipe is superior. I have been validated.

It's not the size of the pipe, it's the optimization you put in that matters

 

[Mr Moose]

There are many games that have shown a 2x advantage in favor of XBO X.

But it doesn't matter, the fact that XBO X doesn't have the same or better fps performance than PSPro is an example of bad optimization. A good optimization on XBO X should achieve the same or better fps than on PSPro and surely it would still have room to present a higher resolution.

OK so... You want it to perform better while also having twice the resolution? PS4 Pros highest res = One X lowest... But you... You want it like this and to have the lead in fps? With that shitty CPU too? The difference isn't even big between them, it's like 1.5fps average or some shit. I think you're expecting too much.

 

[TheRedRiders]

Not surprised honestly. I've been saying it since before launch, it's gonna go back-and-forth.

As to why... honestly, its probably as "simple" as some engines just work better with one machine over the other. This time PS5's smaller faster pipes work better than the XSX's fatter slower pipes, I make this assumption based on all the tech talk I've seen about the machines on here.

It's not accurate to describe the pipe as "fatter", it dismisses the nature and advantage of the hardware design. Slower but wider is a better description for the Series X, and narrower but faster is a good way to describe the PS5.

 

[Lysandros]

I would be curious to see this as well.

I would have thought UE5 would perform better on Series X compared to the PS5 because the engine is compute heavy (which favours the Series X). Yet the results we got for the Matrix demo show almost equal parity between the two systems.

Do we know to which degree UE5 performance depends from things like faster/higher bandwidth GPU caches? As far as i know it still uses hardware rasterization in places which is faster on PS5. And there are the culling/ASYNC side of things, the Geometry Engines, cache scrubbers etc. Loads of unknowns/possibilities. Besides, rated theoretical compute ceiling and the real world world throughput of it can differ quite a bit.

 

[TheRedRiders]

Do we know to which degree UE5 performance depends from things like faster/higher bandwidth GPU caches? As far as i know it still uses hardware rasterization in places which is faster on PS5. And there are the culling/ASYNC side of things, the Geometry Engines, cache scrubbers etc. Loads of unknowns/possibilities. Besides, rated theoretical compute ceiling and the real world world throughput of it can differ quite a bit.

Nanite relies heavily on compute shaders on both consoles, and then it relies on Mesh/Primitive shaders if the compute shaders are not fast enough. Regardless, this would would in theory favour Series X "wider" design, more compute units means more shaders.

UE5 developer mentioned they mainly used a software rasteriser, and that it could beat the fixed function hardware by 3x.

But again, this is all in just theory. As you said there's many unknowns and the fact that the the consoles are performing the same means there could be more going on with the hardware and software. Like other posters have mentioned, the Series X DX12 API causes bottlenecks in RT implementation vs. PS5.

 

[MarkMe2525]

With every post of yours, I'm more and more certain you are some sort of bot. I have so many questions, but sadly you can't answer them.

His posts are always really bizarre and have always stuck out to me. I always imagined English is this guy's 3rd language or something, and that's why his statements are always a little off. This particular one isn't to bad in comparison to some of his others.

 

[Darsxx82]

OK so... You want it to perform better while also having twice the resolution? PS4 Pros highest res = One X lowest... But you... You want it like this and to have the lead in fps? With that shitty CPU too? The difference isn't even big between them, it's like 1.5fps average or some shit. I think you're expecting too much.

You have not understood. I have not asked for 2x resolution and the same fps (although it is a situation that has occurred in many games).

I ask for a GOOD optimization. And good optimization is not what you see on XBO X latest COD games, it's quite the opposite.
A good optimization on XBO X would give the same or better fps and higher resolution. Not 2x than PRo, but a higher resolution at the same fps.

The result on XBO X is an example of inattention and poor optimization. Innecesaary 2x resolution and a mediocre framerate and also with tearing is an example of bad optimization.

 

[Lysandros]

Nanite relies heavily on compute shaders on both consoles, and then it relies on Mesh/Primitive shaders if the compute shaders are not fast enough. Regardless, this would would in theory favour Series X "wider" design, more compute units means more shaders.

Yes more compute units with substantially less L1 cache amount and bandwidth available per CU with slower shader & compute scheduling. As to primitive/mesh shaders the enabler hardware block is the GE which happens to be faster on PS5. But again, i agree, theories and unknowns, we shall see.

 

[GHG]

Where is @Riky these days?

Decommissioned.

Surplus to requirements after VRR was enabled on the PS5.

 

[Caio]

Was banned a few weeks ago and never returned.

No account suicide thread or nothing.

At least he isn't replicating.

I didn't see him anymore in the face-off Threads, and I was wondering what happened to him.

 

[Zathalus]

Honestly the difference in 120 FPS mode appears to be larger than i thought, PS5 is up to 10 FPS ahead on at points on top of running with full Vsync compared to XSX. There some are reports of TAA having higher precision on PS5 with less ghosting based on the analysis, to be confirmed. Series S uses an even lower implementation apparently it is obvious in the video, i didn't hear Tom mentioning it.

It's not that large a difference, at 120 FPS it works out to less than 10%. If it was 30 FPS than 10 FPS would be very significant.

 

[AnotherOne]

 

[DenchDeckard]

Sounds like the series consoles are behind on patches here. What's with the crashes and strange framerate networking issues etc? Enemies jumping around? Going to be one to keep an eye on to see how things go with patches.

 

[Lysandros]

It's not accurate to describe the pipe as "fatter", it dismisses the nature and advantage of the hardware design. Slower but wider is a better description for the Series X, and narrower but faster is a good way to describe the PS5.

I don't find the terms "wider and narrower" quite accurate technically based on architectures. Both are 2SE/4SA designs with same number of rasterizers, prim units etc. XSX has just 4 more CUs sharing the same 128kb of L1 cache per array while PS5 has more ROP hardware. It can be said that they are as wide as each other.

 

[MasterCornholio]

Where is @Riky these days?

Probably embarrassed is what I'm guessing after PS5 got VRR.

 

[TheRedRiders]

I don't find the terms "wider and narrower" quite accurate technically based on architectures. Both are 2SE/4SA designs with same number of rasterizers, prim units etc. XSX has just 4 more CUs sharing the same 128kb of L1 cache per array while PS5 has more ROP hardware. It can be said that they are as wide as each other.

If you look at it from that angle then sure.

I’m speaking strictly in terms of CU count and clock speeds.

 

[avin]

I'm a little puzzled as to why VGTech doesn't see this; the discrepancy in the 120 fps mode, although relatively small, seems too large to have missed based on DF's description. I suppose it's possible DF and VGTech looked at differently patched versions, but DF also seems to have done something unusual, which is measure framerate while spectating in an attempt to capture like for like. I'm not sure I understand exactly what that means, or whether it's a valid way to do this.

avin

 

[elliot5]

Played Warzone on my Series S last night. Looks fine and I don’t think I noticed any frame drops. Game is buggy though in general.

 

[twilo99]

Maybe ms wants to buy the whole thing so they can fix the dev errors on PC..

 

[rickybooby87]

Good perf on PS5, thanks Phil.

 

[coffinbirth]

https://advances.realtimerendering.com/s2022/index.html

It's the "Ray Tracing Open Worlds in UE5" PPT presentation about 2/3 down the page. Discussion on console divergence begins on pg. 62 "Inline Ray Tracing". They speak about how they needed to optimize Series consoles to overcome inefficient DXR ray traversal. Again, not console warring, just giving an example of how one console's performance advantage over the other (whether PS5 or Series X) will often be determined by how better suited they are to a particular engine.

That's actually not what it says at all, lol. That is regarding inline raytracing with DXR 1.0 for PC vs. 1.1 with Series X, where it actually shows GAINS. The optimizations are from the jump to 1.0 for PC to 1.1 with rdna2 for Series X , not regarding Vulkan, ffs.

This was NOT an API comparison with PS5.

And that all goes without the fact that RT is nearly identical across both consoles. Or, conversely, look at an actual UE game with RT with downgraded RT effects on PS5 vs. XSX, such as The Medium.

In reality, DXR is actually more mature than Vulkan regarding efficiency atm. Don't spread misinformation.

 

[Justin9mm]

There are so many variables to consider when comparing XSX vs PS5 side by side game performance like the engine but it just goes to show that whilst Xbox went for raw on paper power, Cerny wasn't talking shit about the design and efficiency of the PS5 to allow devs to squeeze the most out of it!

I have both consoles so not console warring here but I remember when everyone was saying that the extra flops on XSX were going to create a wide gap between the two and dominate but still yet to really see it.

 

[SenjutsuSage]

https://advances.realtimerendering.com/s2022/index.html

It's the "Ray Tracing Open Worlds in UE5" PPT presentation about 2/3 down the page. Discussion on console divergence begins on pg. 62 "Inline Ray Tracing". They speak about how they needed to optimize Series consoles to overcome inefficient DXR ray traversal. Again, not console warring, just giving an example of how one console's performance advantage over the other (whether PS5 or Series X) will often be determined by how better suited they are to a particular engine.

Unless I missed something, at no stage does it actually claim the PS5 has any advantage over the Xbox Series X in the area cited? In fact, I don't see a single mention of the Playstation 5 anywhere in the document. The only mention seen is of DXR 1.0 leaving performance on the table, and how when using In-Line Ray Tracing the Series X saw a performance gain of between 0.3 - 1.7ms.

They cite that DXR 1.0 leaves performance on the table and isn't as inefficient.

Except Series X supports a more advanced tier of DXR called DXR Tier 1.1, which supports the very in-line ray tracing method utilized to bring additional performance in the UE5 document.

I'm guessing you saw some youtube video that literally took an opportunity to spread misinformation about what this info was saying?

DirectX 12 Ultimate solves the issue they were having in this document. It has a higher tier of DXR support. Maybe in the actual video or accompanying audio they mention PS5? But there's certainly no mention of PS5 in the powerpoint. They were highlighting a more advanced form of RT that Series X literally supports in DirectX 12 Ultimate. They were comparing to a less flexible form of DXR. Version 1.0.

https://devblogs.microsoft.com/directx/announcing-directx-12-ultimate/

In other words, there are cases where dynamic-shading ray tracing is better, and there are cases where in-line ray tracing is better. They can even be combined together.

 

[onQ123]

Higher fillrate & triangle rate on PS5

XBO X is also downright poorly optimized on all these recent COD games (on SP and MP modes). It shows quite disappointing performance against PS4Pro....

I think it's more optimization than an engine thing.

PS4 Pro also has the ROP advantage over XBO X like PS5 has a ROP advantage over XBOX Series X

PS4 Pro just had a lot more bottlenecks like the lower memory & slower bandwidth

 

[Max_Po]

ewwww EVIL SONY getting 10 fps+ advantage ...


We know Mark Cerny paniced at last minute and overclocked 8.9 tflop sissy ps5 to bring closer to 12 tflop superior Xbox Ser X ... .


Ewww PS5 is only RDNA1 and my Grandma has a faster SSD to scratch her squared ass crack ....

 

[coffinbirth]

Unless I missed something, at no stage does it actually claim the PS5 has any advantage over the Xbox Series X in the area cited? In fact, I don't see a single mention of the Playstation 5 anywhere in the document. The only mention seen is of DXR 1.0 leaving performance on the table, and how when using In-Line Ray Tracing the Series X saw a performance gain of between 0.3 - 1.7ms.

They cite that DXR 1.0 leaves performance on the table and isn't as inefficient.

Except Series X supports a more advanced tier of DXR called DXR Tier 1.1, which supports the very in-line ray tracing method utilized to bring additional performance in the UE5 document.

I'm guessing you saw some youtube video that literally took an opportunity to spread misinformation about what this info was saying?

DirectX 12 Ultimate solves the issue they were having in this document. It has a higher tier of DXR support. Maybe in the actual video or accompanying audio they mention PS5? But there's certainly no mention of PS5 in the powerpoint. They were highlighting a more advanced form of RT that Series X literally supports in DirectX 12 Ultimate. They were comparing to a less flexible form of DXR. Version 1.0.

https://devblogs.microsoft.com/directx/announcing-directx-12-ultimate/

In other words, there are cases where dynamic-shading ray tracing is better, and there are cases where in-line ray tracing is better. They can even be combined together.

You didn't...because it's not there. Not sure what OP thought they were reading or why they were reading it to begin with if they didn't know what it even meant. Seems to me like they were warring with zero knowledge on the subject, despite claiming otherwise, lol.

 

[adamsapple]

ewwww EVIL SONY getting 10 fps+ advantage ...


We know Mark Cerny paniced at last minute and overclocked 8.9 tflop sissy ps5 to bring closer to 12 tflop superior Xbox Ser X ... .


Ewww PS5 is only RDNA1 and my Grandma has a faster SSD to scratch her squared ass crack ....

It's not even Friday yet ...

 

[Danknugz]

Huh? You literally posted quotes confirming everything that I said. PS5 RT was faster/more efficient than Series consoles.



The "Specialized State Object" for Series optimization is to reduce this inefficiency. It clearly didn't exceed PS5 because otherwise they would have optimized for PS5 as well. Do you need them to explicitly state that "PS5 was still faster even after all of these optimizations"? They are a 3rd party engine developer; that will never, and should never, happen.

I'm not a graphics programmer but you sound like you're reaching. Are you intimately familiar enough with how this works to know what exactly the overhead would be with the different approach with series x to make a meaningful difference? If so that would be more helpful than just pointing out what the slides say.