Support NeoGAF

[Topher]

They matter on PC!

lol....sarcasm?

 

[Lysandros]

So DF is doing a self own? Acting like they weren’t pushing the lie. You mean PS5 having twice the I/O throughput is an important factor?!?!? Who would have thought?!?!

The Great Dictator.

 

[MisterXDTV]

I'm sure when the PS5 Pro is officially announced teraflops will become relevant again...

For DF it's the opposite: they make a big deal about teraflops if Microsoft has more...

When PS4 had more than Xbox One, they stopped doing comparisons

The same will happen if PS5 Pro comes to the market and Xbox stays with Series X

 

[Bogroll]

Teraflops is and always has been a poor metric to compare performance. Not sure how this is new revelation from the DF team.

No it hasn't when comparing the same graphic card architecture before this gen.

 

[hinch7]

No it hasn't when comparing the same graphic card architecture before this gen.

Totally different systems and configurations and API's and yes they using the same generation of parts. There's also different memory configurations plus custom silicon in each one also makes a difference.

Again, using TF is a poor metric to measure performance.

 

[Tams]

Is this the anti-Xbox thread of the day? It grows tiresome.

It's too easy. Xbox give so much ammunition!

 

[Bogroll]

Totally different systems and configurations and API's and yes they using the same generation of parts. There's also different memory configurations plus custom silicon in each one also makes a difference.

Again, using TF is a poor metric to measure performance.

You said has always been. And I said before this gen.

 

[Jibran#one]

No. remember many people buy these consoles for third party games. Having the competition run games much better will hurt sales. just look at 360 vs ps3 a huge reason, 360 was able to compete was cause most mutiplatform games ran better and the biggest Seller was COD

360 was the console for multiplayer games
Back than halo was god tier , gears of war, yearly Forza, pgr .
Trust me horse power wasn't the reason of 360 being popular it was the games

 

[PaintTinJr]

I didn't know about this. Can i have a link for it? Speaking of UE5 we certainly need more benchmarks. I am really curious about why credible outlets didn't make a comparison analysis on games like Layers of Fear using this engine, resolution/FPS findings could be useful.

This was the demo I was remembering

Technical info starts at about 9mins 30 in.

But I think it was the nanite kit-bashing in the making of the AlphaPoint Demo, and not just nanite itself that had the bottleneck, because the Alphapoint demo using nanite with 100Million polys at 1080p30 does hit 2ms, but not like the PS5 doing kit-bashing of megascans (16 Billion polys @1404p locked to 30fps). in 2ms and the XsX was falling back to decal 2D techniques for flooring, etc, which me and onesvenus discussed in the thread at the time around page 3.

https://neogaf.com/threads/the-coalition-alpha-point-technical-demo-and-character-rendering-test-on-unreal-engine-5.1614111/page-2

 

[PaintTinJr]

I love VR, but we can not compare a game like call of the mountain vs the Xbox here. we are going to need actual games on both and for Unreal engine 5 we currently have Remnant 2 and Fortnite. They are more or less identical with a resolution advantage to xbox in fortnite.

But that is the catch 22, the Series consoles can't do the kit bashing with megascans to render at that geometric detail, so we end up with halfway house looking nanite in all faceoffs. Do you not see the problem that always favours bring everything down to the XsX's less effective design?

I know hardware is only part of the equation, because if you just compared lifeless distance rendering via sights - probably static cube maps - in Halo Infinite against fully busy TotK sights rendering, you'd wrongly conclude that the Switch is more powerful than the series, but at some point you do have to take the technical info from each set of devs as a valid point of comparison to efficiency IMO.

 

[killatopak]

Penis size, like always.

7 in. length 1.5 in. girth 90 thrusts per minute. Waiting for new *toys* old ones seem to underperform.

6 in. length 2 in. girth 120 thrusts per minute. Reliable *toys*

If anyone is wondering, mine is 12 inches. It's gaf standard.

 

[zeldaring]

360 was the console for multiplayer games
Back than halo was god tier , gears of war, yearly Forza, pgr .
Trust me horse power wasn't the reason of 360 being popular it was the games

Halo was huge and that's about it. The other games sold well but you are ignoring by far that the biggest franchise cod in that era by far had a huge advantage on 360, and for muliplayer that was huge. Also skyrim ran like shit on ps3 another massive franchise. As I said before it's just not about the games it's a combination of everything. Appeal of the hardware, brand, and public perception. Wiiu had many of the games that moved switch, and mega bombed.

 

[Operations66]

It's too easy. Xbox give so much ammunition!

Not really. It’s still a great system. As of today and for the last couple of years, Xbox Series X is the most powerful video game console ever with a ton of great games including many upgraded BC games, a great controller, and even easy emulation of old systems with dev mode.

There’s a lot of weird hatred here toward Xbox. It’s not my problem, though. Haters choke on their own hate.

Not much more to say about it. Have a good day!

 

[Killjoy-NL]

I don't believe the PS5 is particularly special, no. The primary bottleneck for both systems is the CPU design.

The belief that the PS5 is performing better than its specs is just a case of many not understanding how similar these two GPUs are to each other. They try to compare to PC parts with a similar TF difference, but in the PC world all aspects of the GPU typically move upward with each improvement. On paper both of these console GPUs have distinct statistical advantages, the fact they perform similarly is the expected result for them.

The reality is that MS created the smaller more efficient system and hit their performance target. Sony did likely create the cheaper to produce system which from a business perspective is a big win for them.

Iirc the issue was never about performing above specs, but about efficiency.

And going by Cerny's Road to PS5, they took a different design-philosophy by making PS5 drawing maximum power at all times and shifting power between CPU and GPU depending on which tasks required more energy for maximum performance.

Hence why PS5 seems to perform "above it's specs", when in reality it is designed to minimize bottlenecks so efficiency is closer to 100% than Series X.

Edit:

Obviously, there is the I/O design in PS5 as well, but as far as I can follow tech-talk, it's the power-shift between CPU and GPU that makes the biggest difference.

 

[Tripolygon]

This is thread is really exhausting, from the TF doesn't mean anything to TF is not important so TF is useless. The fact of the matter is TF is not useless and means something. TF is a unit of measurement like any unit of measurement. TF is defined as a trillion float-point operation and is a simple math equation.

Given this number of compute units, how many operations can you do per clock (MHz), multiplied by 2 operations per cycle. This is just a straightforward math, if you use all the CUs to do work how many operations can you do assuming 100% utilization of the ALUs. Because of the nature of software, it means you will never see 100% utilization of ALUs in games (around 50 - 70%) before you become memory bound. That does not mean TF is a useless unit of measurement.

Since we like car analogies. We use horsepower to often discuss how powerful a car is. A horsepower is a simple math equation Torque x RPM / 5,252. You will never achieve 100% utilization of said horsepower when driving because there are various other factors that contribute to losses like friction and heat. It doesn't mean horsepower as a unit of measurement is useless. It is a simple math equation.

 

[sendit]

Not really. It’s still a great system. As of today and for the last couple of years, Xbox Series X is the most powerful video game console ever with a ton of great games including many upgraded BC games, a great controller, and even easy emulation of old systems with dev mode.

There’s a lot of weird hatred here toward Xbox. It’s not my problem, though. Haters choke on their own hate.

Not much more to say about it. Have a good day!

You maybe the sole person in 2023 that still believes that. Teraflops isn't everything.

 

[Lysandros]

This was the demo I was remembering

Technical info starts at about 9mins 30 in.

But I think it was the nanite kit-bashing in the making of the AlphaPoint Demo, and not just nanite itself that had the bottleneck, because the Alphapoint demo using nanite with 100Million polys at 1080p30 does hit 2ms, but not like the PS5 doing kit-bashing of megascans (16 Billion polys @1404p locked to 30fps). in 2ms and the XsX was falling back to decal 2D techniques for flooring, etc, which me and onesvenus discussed in the thread at the time around page 3.

https://neogaf.com/threads/the-coalition-alpha-point-technical-demo-and-character-rendering-test-on-unreal-engine-5.1614111/page-2

I see, thanks a lot.

 

[ShadowkillXNA]

I don't know what to tell you @Mr.Phoenix, if you think the XSX is out of touch with the current console market. If that's the case, well you know what that says about the PS5.

You do understand the ps5 has outsold the xbox series by 40 million to 21 million to date right?
The ps5 is not in any way, shape or form out of touch with the console market lmfao.

 

[MasterCornholio]

I don't know what to tell you @Mr.Phoenix, if you think the XSX is out of touch with the current console market. If that's the case, well you know what that says about the PS5.

Sony and MS released $500 consoles (the disk drive models) at launch and both boxes perform similarly to each other. If MS needs a couple more TF to make that happen due to differences in the way the APIs and SDKs are structured, it was smart for them to pick those up in the spec. With the specs they chose they can reach performance parity (often they do get a 15-20% edge in resolution which is all the GPU difference amounts to, by the way) and still get the long-term benefits that their more hardware agnostic API approach provides. Mission accomplished.

It just says that more people want PS5 than Xbox Series. That's all the market is saying at the moment.

You can intepret that how you wish though.

 

[twilo99]

The same will happen if PS5 Pro comes to the market and Xbox stays with Series X

Will the difference between pro and xsx be the same as its between the ps5 and xsx, namely a few flops?

 

[MasterCornholio]

Will the difference between pro and xsx be the same as its between the ps5 and xsx, namely a few flops?

There was a pretty big memory difference between the PS4 Pro and the Xbox One X.

8GB vs 12GB.

 

[Lysandros]

This is thread is really exhausting, from the TF doesn't mean anything to TF is not important so TF is useless. The fact of the matter is TF is not useless and means something. TF is a unit of measurement like any unit of measurement. TF is defined as a trillion float-point operation and is a simple math equation.

Given this number of compute units, how many operations can you do per clock (MHz), multiplied by 2 operations per cycle. This is just a straightforward math, if you use all the CUs to do work how many operations can you do assuming 100% utilization of the ALUs. Because of the nature of software, it means you will never see 100% utilization of ALUs in games (around 50 - 70%) before you become memory bound. That does not mean TF is a useless unit of measurement.

Since we like car analogies. We use horsepower to often discuss how powerful a car is. A horsepower is a simple math equation Torque x RPM / 5,252. You will never achieve 100% utilization of said horsepower when driving because there are various other factors that contribute to losses like friction and heat. It doesn't mean horsepower as a unit of measurement is useless. It is a simple math equation.

In that case the engine's horse power isn't solely consisting of floating point operations per second thought, that is the important distinction. It's only a quite important 'piece' of the puzzle. As to compute efficiency/CU saturation, it is also dictated by the hardware layout/architecture besides the software.

 

[twilo99]

There was a pretty big memory difference between the PS4 Pro and the Xbox One X.

I expect there to be a substantial upgrade all around, otherwise doing a “pro” version makes little sense to me

 

[hinch7]

You said has always been.

There's more nuance to that answer than saying everything is black and white. But measuring performance based on one metric on a GPU is meaningless as it also depends on the total system setup which has a lot of different variables things in the hardware itself. And that doesn't include software implementation. You could have the best silicon in the world and match with with some poor memory configuration (for example) and it would drastically affect its performance. Or lets say you had one GPU with higher core count and but less VRAM and lower core GPU with higher clocks but more RAM.. The former would tank in performance once it goes over that memory threshold. But both will perform differently in certain scenarios. And thats just a couple examples. Then ofc you have different architectures and you get my point.

 

[MasterCornholio]

I expect there to be a substantial upgrade all around, otherwise doing a “pro” version makes little sense to me

Depending on how Microsoft and Sony design their Pro consoles the differences between the two can be more than just a few flops. Like how the One X and Pro had differences beyond the TF count.

 

[MisterXDTV]

Depending on how Microsoft and Sony design their Pro consoles the differences between the two can be more than just a few flops. Like how the One X and Pro had differences beyond the TF count.

The One X was much more than a mid-gen, Microsoft was ashamed of the original Xbox One and they changed everything but the CPU (because there wasn't an alternative to Jaguar in 2017)

Then 2 years and a half later they told people that Series S was a better console...

Anyway I don't see why MS should make a THIRD console this gen, they are not even selling the two they have on the market right now and they seem to care about PC more than consoles

 

[Bogroll]

There's more nuance to that answer than saying everything is black and white. But measuring performance based on one metric on a GPU is meaningless as it depends on the total system setup and a lot of different variables things in hardware itself and not just one aspect of it and that doesn't include software implementation. You could have the best silicon in the world and match with with some poor memory configuration (for example) and it would drastically affect its performance. Or lets say you had one GPU with lesser memory.. and core count and clocked GPU, and one lesser with more VRAM and performance would tank once you go over that threshold on the former because it goes over that memory threshold. And thats just a couple examples. Then ofc you have different architectures and you get my point.

Yes of course. But you claim TF has always been useless. Show me a Graphic card with less TF with same architecture (other than current consoles) that outperforms the higher TF card.

I'm not saying there isn't but got to be rare. I'd guess 3060 12gb v 3060ti 8gb.

 

[Tripolygon]

In that case the engine's horse power isn't solely consisting of floating point operations per second thought, that is the important distinction. It's only a quite important 'piece' of the puzzle. As to compute efficiency/CU saturation, it is also dictated by the hardware layout/architecture besides the software.

That goes without saying, the overall architecture affects how your system performs including external factors like ambient temperature, which is the point of the analogy, there are factors that would affect 100% utilization CUs but a floating point is a floating point is a floating point. It's just a simple math equation and it is the same on Intel, AMD, Nvidia, Imagination Technologies, Qualcomm etc. A terrible driver (pun not intended) can greatly affect how efficient a GPU operates by the same token a bad driver can affect how effective you use the horsepower your engine generates. It's just an analogy not a direct comparison.

 

[DaGwaphics]

It just says that more people want PS5 than Xbox Series. That's all the market is saying at the moment.

You can intepret that how you wish though.

You do understand the ps5 has outsold the xbox series by 40 million to 21 million to date right?
The ps5 is not in any way, shape or form out of touch with the console market lmfao.

That's some extremely pathetic console warring right there. LOL

What do sales have to do with the consoles on a hardware level? The systems are so similar that if the hardware of one is "out of touch" the other is guaranteed to be in the same position. Better example of hardware that was "out of touch" with the state of the market was the Saturn, where they misunderstood the importance of 3D at the time. Maybe if MS had gone with a mechanical HDD you could say something like that.

 

[onQ123]

Amen to that.

I said ps5 pro "is" around 22-23tf in a sense that it performance like a machine that has a 22-23tf gpu, not that the actual count is 22-23tf, in the same way that a 10.28 tf machine (ps5) performances like a 12tf (XSX) or even better

Stop using the silly Tflops metrics once and for all!!

By saying that you're doing exactly what you said we shouldn't do lol.

PS5 does not perform like a 12tf machine it performs like a 10tf machine that's clocked high for better fixed function operations & boosted memory subsystem.


I think the smart thing to do when creating consoles is to use compute when you don't know where the industry is headed but use fixed function when you do know what's going to be used more in games.

A pro console would actually be a good time to use Ray-Tracing hardware .

 

[EddieVanHelgen]

Let's get some Halo Infinite quality visuals on PS5 before we take this too seriously.

 

[JackMcGunns]

If only the series s only problem was teraflops.

Series S improved over Xbox One X in every aspect except for teraflops, yet everyone complained, seems pretty hypocritical.

 

[DaGwaphics]

Obviously, there is the I/O design in PS5 as well, but as far as I can follow tech-talk, it's the power-shift between CPU and GPU that makes the biggest difference.

The shift between CPU and GPU doesn't affect the maximum capabilities of either the CPU or the GPU at all. The peak specs are based on the maximum clocks. Thus the statistical comparisons are based on the highest possible values for both.

The bigger difference between the systems is the APIs and the level of hardware abstraction that exists on the platforms. If MS had gone with a 10TF GPU they would have ended up under powered. But again, in many cases the XSX gets the pixel advantage that the differences between the GPUs would suggest anyway. At the end of the day these are consoles that launched at the same price and mostly perform the same.

 

[Lysandros]

By saying that you're doing exactly what you said we shouldn't do lol.

PS5 does not perform like a 12tf machine it performs like a 10tf machine that's clocked high for better fixed function operations & boosted memory subsystem.


I think the smart thing to do when creating consoles is to use compute when you don't know where the industry is headed but use fixed function when you do know what's going to be used more in games.

A pro console would actually be a good time to use Ray-Tracing hardware .

In other words there is not such a thing as "performing like a X. Teraflops machine" in absolute terms. The statement in itself is mostly meaningless.

 

[64bitmodels]

Is this the anti-Xbox thread of the day? It grows tiresome.

 

[MisterXDTV]

Series S improved over Xbox One X in every aspect except for teraflops, yet everyone complained, seems pretty hypocritical.

What about memory???

 

[SolidQ]

 

[Killjoy-NL]

The shift between CPU and GPU doesn't affect the maximum capabilities of either the CPU or the GPU at all. The peak specs are based on the maximum clocks. Thus the statistical comparisons are based on the highest possible values for both.

The bigger difference between the systems is the APIs and the level of hardware abstraction that exists on the platforms. If MS had gone with a 10TF GPU they would have ended up under powered. But again, in many cases the XSX gets the pixel advantage that the differences between the GPUs would suggest anyway. At the end of the day these are consoles that launched at the same price and mostly perform the same.

I never said the shift affected maximum capabilities, nor did anyone else.

But as you said, comparisons are based on highest possible values.

If one platform has less Tflops, but operates at max performance all the time (theoretically), while the one with more Tflops consistently struggles to maintain max performance, it all evens out.

This was the design philosophy behind PS5 and we've seen the real-world results, where PS5 and Xbox go toe-to-toe.

The reason Xbox went with a more powerful GPU was because of all the bottlenecks, which we see hampering Series X performance vs PS5.
They went with a conventional design. It doesn't make them smart at all. They just played it save.

And now we see people like DF who downplayed PS5 pre-release come to realize what Cerny had been saying during Road to Playstation.

That's what I got from all the discussions.

 

[Gaiff]

For DF it's the opposite: they make a big deal about teraflops if Microsoft has more...

When PS4 had more than Xbox One, they stopped doing comparisons

The same will happen if PS5 Pro comes to the market and Xbox stays with Series X

Wait, what? DF spent the entire 8th gen showing the PS4 beating the Xbox One and kept bringing up TFLOPs over and over again which is what led them to hold onto the belief that the Series X would win out against the PS5.

I know some of you have a bizarre hate-boner for DF but at least don't spout bullshit that is easily disprovable.

 

[Killjoy-NL]

Wait, what? DF spent the entire 8th gen showing the PS4 beating the Xbox One and kept bringing up TFLOPs over and over again which is what led them to hold onto the belief that the Series X would win out against the PS5.

I know some of you have a bizarre hate-boner for DF but at least don't spout bullshit that is easily disprovable.

Well, at some point, you can't deny reality.
Even DF knows that.

 

[Freelance Yakuza]

Series S improved over Xbox One X in every aspect except for teraflops, yet everyone complained, seems pretty hypocritical.

One X = 12 GB RAM
Series S = 10 GB RAM

 

[Vox Machina]

lol....sarcasm?

No? PC games usually don't cap framerates, so any incremental improvement in compute should generally lead to increased performance.

Consoles capping framerates at 30 or 60 means any performance over those caps is just wasted.

 

[Bogroll]

Wait, what? DF spent the entire 8th gen showing the PS4 beating the Xbox One and kept bringing up TFLOPs over and over again which is what led them to hold onto the belief that the Series X would win out against the PS5.

I know some of you have a bizarre hate-boner for DF but at least don't spout bullshit that is easily disprovable.

I just don't know what planet that poster has been living on.

 

[MasterCornholio]

That's some extremely pathetic console warring right there. LOL

What do sales have to do with the consoles on a hardware level? The systems are so similar that if the hardware of one is "out of touch" the other is guaranteed to be in the same position. Better example of hardware that was "out of touch" with the state of the market was the Saturn, where they misunderstood the importance of 3D at the time. Maybe if MS had gone with a mechanical HDD you could say something like that.

Well you talked about being out of touch with the console market. It's why sales was brought into the conversation because thst indicates market trends.

Maybe don't mention the market next time if your focus is on the hardware.

 

[Topher]

No? PC games usually don't cap framerates, so any incremental improvement in compute should generally lead to increased performance.

Consoles capping framerates at 30 or 60 means any performance over those caps is just wasted.

Ah.....ok. It sounded like you were saying TF matter to PCs in the same way they have "mattered" in console war madness. I get that's not what you meant now.

 

[Kataploom]

There is absolutely not such an established fact or general rule. PS5 can well outperform XSX in GPU bound scenarios since it has as many hardware advantages on that front. Notably in (pixel) fill rate, local bandwidth and geometry bound scenarios PS5 should be expected to outperform XSX. There are plentiful of cases where PS5 has even the resolution advantage. We should stop acting as if XSX GPU is all around superior to PS5's at some point, this is a plain faulty assumption.

And as far as I'm concerned i have nothing bad to say about XSX. It has undeniable advantages such as power consumption and form factor and it's certainly competitive with PS5 in performance front.

Not saying it shouldn't, just pointing out the evidence I've seen. Flops are a very "liquid" measure in the sense it depends on different stuff being measured so one thing can add more than others depending on devices, but most of the time for GPU bound scenarios I have seen Xbox performing a little better than ps5. At least on digital foundry analysis.

 

[Vox Machina]

Ah.....ok. It sounded like you were saying TF matter to PCs in the same way they have "mattered" in console war madness. I get that's not what you meant now.

Nah, I do like that some games are shipping with unlocked framerate modes on console now though, although they seem to be few and far between.

 

[SlimySnake]

They aren’t really upping the tflops though. Increased tflops is just a byproduct of a more powerful system through wattage increase/more cores/higher efficiency/superior tech

Tflops arent a byproduct of wattage increase. Wattage is a byproduct of increased tflops. You have more Cores and higher clocks which means more power usage. By your logic, the PS3 which was also a 200 watt console is just as powerful as a PS5 which has 44x more tflops just because it had the same wattage.

They are increasing the CUs which in turn increases the tflops. This after Cerny made excuses to why he was sticking with 36 CUs. I wonder why. Now that he is increasing CUs, tflops dont matter. Sure.

Unbelievable what has happened to online discourse since last gen with Cerny's damage control making half of the online userbase downplaying tflops and Xbox's insane boardroom design decisions to create an albatross that loses to the PS5 despite having tflops. Even though in the PC space, a fucking 10 tflops rdna 2.0 card would never get beaten by a 12 tflosp rdna 2.0 card.

 

[Bojji]

Series S improved over Xbox One X in every aspect except for teraflops, yet everyone complained, seems pretty hypocritical.

Series S has better CPU and storage but weaker GPU, less RAM and less memory bandwidth.

 

[Kataploom]

How would a once free ability to play online becoming paid be an 'optional service'? That's not optional.

You don't pay for it. That's it. Just play offline. Also PC didn't require that.