• 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.

Gavin Stevens

Formerly 'o'dium'
Ahh ok, thanks dude.

It really is a case of not wanting other chips to be different. I have extreme examples, but let’s say you have 56 compute units. You won’t use all 56 because the yield chance is too low, so you use less, let’s say 52. That leaves 4 to play with, so if you have any more than 4 faulty, you bin the chip and take the loss (or use it for something else). But by making it 4 and not 0, or 1, you have given yourself a much better chance at having higher yields and thus not burning money.
 

kareemna

Member
It really is a case of not wanting other chips to be different. I have extreme examples, but let’s say you have 56 compute units. You won’t use all 56 because the yield chance is too low, so you use less, let’s say 52. That leaves 4 to play with, so if you have any more than 4 faulty, you bin the chip and take the loss (or use it for something else). But by making it 4 and not 0, or 1, you have given yourself a much better chance at having higher yields and thus not burning money.

Usually the chips with lesser active CUs are sold as lower end GPUs, right?
 

Gavin Stevens

Formerly 'o'dium'
Usually the chips with lesser active CUs are sold as lower end GPUs, right?

Usually they are repurposed yes. However, that still eats into profits, because it was originally forecast for a higher earning. It’s a smart way of getting at lest part of the cost back :) But that’s with GPUs... I’m not sure if the same can be applied so easily on such a massive scale with consoles. I mean, GPUs don’t sell anywhere near as much as consoles do they? (Genuine question, not sure).

It’s probably the reason they picked 7nm over 7nm+ here, because the chips will have higher base yields, and on such a mass production scale, you don’t want to burn money.
 
Last edited:

B_Boss

Member
It really is a case of not wanting other chips to be different. I have extreme examples, but let’s say you have 56 compute units. You won’t use all 56 because the yield chance is too low, so you use less, let’s say 52. That leaves 4 to play with, so if you have any more than 4 faulty, you bin the chip and take the loss (or use it for something else). But by making it 4 and not 0, or 1, you have given yourself a much better chance at having higher yields and thus not burning money.

Is it a given that some will be faulty?
 

kareemna

Member
Usually they are repurposed yes. However, that still eats into profits, because it was originally forecast for a higher earning. It’s a smart way of getting at lest part of the cost back :) But that’s with GPUs... I’m not sure if the same can be applied so easily on such a massive scale with consoles. I mean, GPUs don’t sell anywhere near as much as consoles do they? (Genuine question, not sure).

It’s probably the reason they picked 7nm over 7nm+ here, because the chips will have higher base yields, and on such a mass production scale, you don’t want to burn money.

This is why i mentioned GPUs as with consoles lower yields could not generally be repurposed to be sold in any way for form. 7nm has higher yields indeed, I think over at Beyond3d discussed this.
 

Sosokrates

Report me if I continue to console war
But that's really not true. All PS consoles were strong at their release. They clearly overdid it on the PS3.

The PS3 cannot be compared to a 12TF RDNA2 console today either. The BOM cost of the PS3 was around $ 800 at the beginning. You will NOT reach such BOM costs this gen, even with 12TF@RDNA2.

So, there is absolutely no reason to believe sony relases a clearly worse console than the XSX.

There are 3 reasons the PS1 , PS2 and PS4, and who ever said they were not strong? I certainly did not, what I am saying is that Sony have 3 out of 4 times have tried to found a balance of power and price, 3 out of 4 times they have not been about chasing the power crown at the sacrifice of a great price.
 

Gavin Stevens

Formerly 'o'dium'
Is it a given that some will be faulty?

In any product it’s a given you will have a small percentage of faulty units.

With stuff like this? It’s even more likely. Sadly we have yet to create a process that can make such chips perfectly, not at this small a scale.

I mean even expensive tvs still have dead pixels a lot of the time, and those are quite large.
 

Gediminas

Banned
fTVfkvD.jpg
wait what? do you want to say PS5 does not have RDNA2? that what i thought! #PS5inthedust
 

Sosokrates

Report me if I continue to console war
Too many people saying "bad design" in regards to a 9.2 TFLOP PS5 have no idea what Sony's goals are this upcoming cycle. The original Xbox One wasn't a bad design. It was the perfect design for what the Microsoft leadership at the time thought the direction of the industry would be. They just happened to be dead wrong. Sony just happened to stick with the status quo with the PS4 and ended up being right about the future of the games console. The PS4 wasn't some monster-spec machine when it launched. It just happened to be competing directly with a do-everything cable-box wannabe machine. Don't forget that Sony bought into the same post-iPhone era propaganda as MS, they just had a more modest approach with a modestly specced machine with their fingers and toes dipped in motion control, media center and VR technology. But the PS4 was priced to move.

Sony might have the same approach as last gen. Release a modestly powerful machine priced to move and then release a more powerful revision two or three years later. Microsoft under Phil Spencer has always been about pure horsepower, so the Series X being a monster machine is not out of character. Post-Kutaragi Sony doesn't seem to be about winning spec sheet wars. They make their bread by shipping sleek, modestly powerful machines at a mass market price. Spec wars are for forum fanboys. If Sony has an elegant way of hardware backwards compatibility that doesn't break the bank and can deliver a 9.2 Navi TFLOPS box at $399, they'll jump at the chance and dance all the way to the bank.

A 9.2 TFLOPS (Navi) PS5 is still more than twice as powerful as PS4 Pro and about 10 times more performant than the PS4 which is about what you should expect from a generational leap. It would be typical of Sony to make a move like this.

Don't forget that a Sony console being more powerful than a Microsoft console was a generational anomaly. Xbox was more advanced than PS2 and the Xbox 360 was more powerful than the PS3. Microsoft gambled with TV, TV, TV, and Sony just happened to land in a sweet spot with specs that really didn't stand toe-to-toe with affordable PC hardware that was already out.

TLDR: You don't know what Sony's priorities are. That priority is driving the PS5's design. But you can be sure they're not trying to win a fanboy PR war. If PS5 had a horsepower edge, that would have either been announced or it would have "leaked" by now. My guess is that Sony rightly knows that it's the demos that will knock socks off and drive excitement. They'd rather have actual games to show than do a repeat of the PS3 2005 E3. That takes time. When the PS5 floodgates open it will be heavy and unrelenting up until launch. All they need to do is have their heavy hitters show actual games and 9.2 or 14 or 20 will just be a bunch of numbers.

That's just my opinion. No claim of inside information.
 

Ellery

Member
Sooo, let's get down to figures, shall we.
AMD's +50% perf/watt claim.

5700XT, 9.7TF, 225 watt.
TF/w => 0,0431

RDNA2 => 0,0646
13TF => 201 watt for GPU alone

*Assuming that perf/flop doesn't change significantly.

Two important things to mention here is that consoles run at lower clocks/voltages than Graphic Cards and you are also comparing the TDP of a graphic cards (which is not something you should do) to a possible 13TF RDNA2 Unit. The TDP doesn't mean power usage during certain tasks but is the thermal design power, which is slightly different.

13TF RDNA2 should be much lower than 201 watt. If you factor in lower clocks aswell (which is a given for consoles because they don't have gigantic cooling solutions like graphic cards) to hit the sweetspot then you will end up at lower than 201 W.
 
Too many people saying "bad design" in regards to a 9.2 TFLOP PS5 have no idea what Sony's goals are this upcoming cycle. The original Xbox One wasn't a bad design. It was the perfect design for what the Microsoft leadership at the time thought the direction of the industry would be. They just happened to be dead wrong. Sony just happened to stick with the status quo with the PS4 and ended up being right about the future of the games console. The PS4 wasn't some monster-spec machine when it launched. It just happened to be competing directly with a do-everything cable-box wannabe machine. Don't forget that Sony bought into the same post-iPhone era propaganda as MS, they just had a more modest approach with a modestly specced machine with their fingers and toes dipped in motion control, media center and VR technology. But the PS4 was priced to move.

Sony might have the same approach as last gen. Release a modestly powerful machine priced to move and then release a more powerful revision two or three years later. Microsoft under Phil Spencer has always been about pure horsepower, so the Series X being a monster machine is not out of character. Post-Kutaragi Sony doesn't seem to be about winning spec sheet wars. They make their bread by shipping sleek, modestly powerful machines at a mass market price. Spec wars are for forum fanboys. If Sony has an elegant way of hardware backwards compatibility that doesn't break the bank and can deliver a 9.2 Navi TFLOPS box at $399, they'll jump at the chance and dance all the way to the bank.

A 9.2 TFLOPS (Navi) PS5 is still more than twice as powerful as PS4 Pro and about 10 times more performant than the PS4 which is about what you should expect from a generational leap. It would be typical of Sony to make a move like this.

Don't forget that a Sony console being more powerful than a Microsoft console was a generational anomaly. Xbox was more advanced than PS2 and the Xbox 360 was more powerful than the PS3. Microsoft gambled with TV, TV, TV, and Sony just happened to land in a sweet spot with specs that really didn't stand toe-to-toe with affordable PC hardware that was already out.

TLDR: You don't know what Sony's priorities are. That priority is driving the PS5's design. But you can be sure they're not trying to win a fanboy PR war. If PS5 had a horsepower edge, that would have either been announced or it would have "leaked" by now. My guess is that Sony rightly knows that it's the demos that will knock socks off and drive excitement. They'd rather have actual games to show than do a repeat of the PS3 2005 E3. That takes time. When the PS5 floodgates open it will be heavy and unrelenting up until launch. All they need to do is have their heavy hitters show actual games and 9.2 or 14 or 20 will just be a bunch of numbers.

That's just my opinion. No claim of inside information.
If we consider RDNA 2 into account, those 9.2 TFs are much more impressive than two times PS4 Pro, it would already be around 4-5 times a PS4 Pro as practical power, at least according to AMD data. It's true that this isn't the generation of TFs, average people are gonna waaay underastimate this machines I think.
Actually, there wasn't a quote from Sony saying that PS5 is infact 4 times more powerful than PS4 Pro?
 
Last edited:

Ellery

Member
If we consider RDNA 2 into account, those 9.2 TF are much more impressive than two times PS4 Pro, it would already be around 4-5 times a PS4 Pro as practical power, at least according to AMD data. It's true that this isn't the generation of TF, average people are gonna waaay underastimate this machines I think.
Actually, there wasn't a quote from Sony saying that PS5 is infact 4 times more powerfull than PS4 Pro?

I don't remember any quote like this, but what you say is indeed true and the reason why people should be excited about upcoming consoles.

Well I don't want to speculate too much about how many TF the PS5 is going to have, but if we look at the Xbox Series X being 12 TF RDNA2 then it is truly impressive. It is a gigantic jump over Xbox One X 6 TF. Much much more than 2x the graphical performance.
The same will be true for the PS5 even if the TF will be on the lower side of speculation it will be a gigantic leap over the PS4 Pro. And if rumors are to be believed about the PS5 being somewhere around the Series X then I will have a hard time stopping my drool just thinking about what the talented developers are going to bring to life with such a strong machine.
 

Sosokrates

Report me if I continue to console war
If we consider RDNA 2 into account, those 9.2 TFs are much more impressive than two times PS4 Pro, it would already be around 4-5 times a PS4 Pro as practical power, at least according to AMD data. It's true that this isn't the generation of TFs, average people are gonna waaay underastimate this machines I think.
Actually, there wasn't a quote from Sony saying that PS5 is infact 4 times more powerful than PS4 Pro?

Yes next gen is going to be very exciting regardless of where the specs end up.
 
I don't remember any quote like this, but what you say is indeed true and the reason why people should be excited about upcoming consoles.

Well I don't want to speculate too much about how many TF the PS5 is going to have, but if we look at the Xbox Series X being 12 TF RDNA2 then it is truly impressive. It is a gigantic jump over Xbox One X 6 TF. Much much more than 2x the graphical performance.
The same will be true for the PS5 even if the TF will be on the lower side of speculation it will be a gigantic leap over the PS4 Pro. And if rumors are to be believed about the PS5 being somewhere around the Series X then I will have a hard time stopping my drool just thinking about what the talented developers are going to bring to life with such a strong machine.
Probably was just a rumor, I don't want to spread shit from nothing.
And I know we are going in circles, but we are talking only about GPU power here. Zen 2 and high speed SSD will add up to the structural differences from PS4/Ps4 Pro to the point where PS5 could be enormously superior to those. Taking into account only one thing do not make justice to the leap we are going to see. Without those, 4x the power would be wasted.
 
Last edited:

IkarugaDE

Member
There are 3 reasons the PS1 , PS2 and PS4, and who ever said they were not strong? I certainly did not, what I am saying is that Sony have 3 out of 4 times have tried to found a balance of power and price, 3 out of 4 times they have not been about chasing the power crown at the sacrifice of a great price.
Hey dude, not every statement automatically means that I claim that it was made by someone. :messenger_beaming:

500$ for a about 12TF RDNA2 system definitely IS a balance between power and price. You can't compare the actual situation with any in the past, especially with a competitor like MS with it's 12TF machine. It IS possible sony delivers a 400$ System with fewer TF, but it is not really likely. In addition to the fact that people spend more money on electronics today, a competitive product has to be brought onto the market.
 

Bo_Hazem

Banned
Nvidia just have way more products out there and more features packed into their current cards, but price to performance AMD still beat them at 1080p and 1440p. Nvidia will have to stop playing around cause AMD is coming hard for them with RDNA2.

That's why I support RED

myuHHTt.png
 

geordiemp

Member
Too many people saying "bad design" in regards to a 9.2 TFLOP PS5 have no idea what Sony's goals are this upcoming cycle. The original Xbox One wasn't a bad design. It was the perfect design for what the Microsoft leadership at the time thought the direction of the industry would be. They just happened to be dead wrong. Sony just happened to stick with the status quo with the PS4 and ended up being right about the future of the games console. The PS4 wasn't some monster-spec machine when it launched. It just happened to be competing directly with a do-everything cable-box wannabe machine. Don't forget that Sony bought into the same post-iPhone era propaganda as MS, they just had a more modest approach with a modestly specced machine with their fingers and toes dipped in motion control, media center and VR technology. But the PS4 was priced to move.

Sony might have the same approach as last gen. Release a modestly powerful machine priced to move and then release a more powerful revision two or three years later. Microsoft under Phil Spencer has always been about pure horsepower, so the Series X being a monster machine is not out of character. Post-Kutaragi Sony doesn't seem to be about winning spec sheet wars. They make their bread by shipping sleek, modestly powerful machines at a mass market price. Spec wars are for forum fanboys. If Sony has an elegant way of hardware backwards compatibility that doesn't break the bank and can deliver a 9.2 Navi TFLOPS box at $399, they'll jump at the chance and dance all the way to the bank.

A 9.2 TFLOPS (Navi) PS5 is still more than twice as powerful as PS4 Pro and about 10 times more performant than the PS4 which is about what you should expect from a generational leap. It would be typical of Sony to make a move like this.

Don't forget that a Sony console being more powerful than a Microsoft console was a generational anomaly. Xbox was more advanced than PS2 and the Xbox 360 was more powerful than the PS3. Microsoft gambled with TV, TV, TV, and Sony just happened to land in a sweet spot with specs that really didn't stand toe-to-toe with affordable PC hardware that was already out.

TLDR: You don't know what Sony's priorities are. That priority is driving the PS5's design. But you can be sure they're not trying to win a fanboy PR war. If PS5 had a horsepower edge, that would have either been announced or it would have "leaked" by now. My guess is that Sony rightly knows that it's the demos that will knock socks off and drive excitement. They'd rather have actual games to show than do a repeat of the PS3 2005 E3. That takes time. When the PS5 floodgates open it will be heavy and unrelenting up until launch. All they need to do is have their heavy hitters show actual games and 9.2 or 14 or 20 will just be a bunch of numbers.

That's just my opinion. No claim of inside information.

Problem is, the day one buys and buyers with high budget means that in the first 3-6 months is when you want to sell a pro / high spec meachine. Its year 2 when sales need a low cost machine, everyone knows that.

As such, to launch lower spec and bring higher spec in 2 years makes liittle business sense if the technology si available now, which it is.

Pro and X launched when tech was available, node shrink.
 
I admit that I love the intricacies of the technology we discuss, but I'm a bit of a layman in regards to it all.

Assuming 13TF was always the target, as Osiris has said, knowing the new efficiency numbers, is it more or less realistic than it was before?

No mentions of HBM2, which I'd seen thrown around a few times that it would be neccesary to hit those higher clocks rather than DDR6. I guess because it runs cooler?

Some people hold fast to 14, which I'd like to be an optimist, but think it is a bit far-fetched.

Anyway--grateful that RDNA2 has been confirmed for both consoles and that essentially puts an end to the silly "no RT" thing going around.

I think it'll still be classified as 12TF at the reveal myself.
 

Dunnas

Member
It is just a demonstration, as others have done after me, in mathematical terms that 50% better perf/watt improvements are a 33% performance increase for the same watt.

The way AMD is wording it makes some people believe that a chip which has a gaming performance of 100fps for 100 watt now has 150fps for 100 watt but in reality it would result in 133fps for 100 watt.
No, that is not how the math works. A 50% increase in performance per what does actually mean 100 to 150fps at 100 watts. The 33% figure is the reduction in wattage for the same performance. For example.

RDNA - 100w = 100 fps

RDN2 - 100w = 150 Ffps
RDNA2 - 66.66w = 100 FPS - a 33% reduction in watts for the same fps as RDNA2
 
Status
Not open for further replies.
Top Bottom