Current Generation Hardware Speculation with a Technical Spin [post GDC 2020] [XBSX, PS5]

[AlNom]

Here is it 8MB of L3, there is a big chance this the CPU of the two next generation consoles.

This indicates 12MB for the 4900HS.
https://www.notebookcheck.net/AMD-Ryzen-9-4900HS-Processor-Benchmarks-and-Specs.454860.0.html

Weird. AMD site indicates 8MB.
https://www.amd.com/en/products/apu/amd-ryzen-9-4900hs


Whoops! Maybe they accidentally added the L2 cache (4MB).

 

[see colon]

The naming convention is PS5's biggest advantage. Lol.

Just because the name is clocked higher doesn't mean that it's not limited by it's total lack of X's. XboX Series X has 3 X's. A name so powerful it should be considered pornographic! [insert joke about pushing out LODs]

Have you tried searching for "fastest console" or "most powerful console"?

What about "Point Break"

In all seriousness, if you use Bing for "Most Powerful Console" it brings up results for Xbox One X, no mention of Sony at all, and no mention of future consoles. Probably a bit of a bias in those results, though.

 

[KeanuReeves]

You're mistaken that it had nothing to do with clocks. He even mentions, in a separate section to the one I linked, how they would have had to clock the CPU lower due to the occasions when the new 256-bit AVX instructions were used causing the CPU to draw a lot more power. Fans not ramping out of control is an effect of the new system they are using. It is not the reason that system was developed.

Well, I will accept that the wording may not have been noise, but instead the problem was power draw and fan size instead. That is actually something he said verbatim.

He just laid out their entire approach in deciding frequency clocks for previous gens and next-gen, and because PS5 supports instructions (avx 256bit as you said) that produce adverse amounts of heat they had to come up with a new solution. There is nothing here that says "to lock at 2GHz, we had to...". I don't see that as me being mistaken, but rather the semantics was very slightly different. My point still stands.

The continuous boost solution is not there to reach some GHz goal, but rather to manage power draw, and fan size (as Mark Cerny explicitly said at 34:33).

 

[mrcorbo]

Well, I will accept that the wording may not have been noise, but instead the problem was power draw and fan size instead. That is actually something he said verbatim.

He just laid out their entire approach in deciding frequency clocks for previous gens and next-gen, and because PS5 supports instructions (avx 256bit as you said) that produce adverse amounts of heat they had to come up with a new solution. There is nothing here that says "to lock at 2GHz, we had to...". I don't see that as me being mistaken, but rather the semantics was very slightly different. My point still stands.

The continuous boost solution is not there to reach some GHz goal, but rather to manage power draw, and fan size (as Mark Cerny explicitly said at 34:33).

We really don't know for absolutely certain which between cooling system consistency and performance goals was the chicken or the egg as a driver for the development of their solution. In the end though, it's both. The variable clocks is what allows them to reach the clocks they did with the cooling system and power delivery they implemented.

 

[MrFox]

Well, I will accept that the wording may not have been noise, but instead the problem was power draw and fan size instead. That is actually something he said verbatim.

He just laid out their entire approach in deciding frequency clocks for previous gens and next-gen, and because PS5 supports instructions (avx 256bit as you said) that produce adverse amounts of heat they had to come up with a new solution. There is nothing here that says "to lock at 2GHz, we had to...". I don't see that as me being mistaken, but rather the semantics was very slightly different. My point still stands.

The continuous boost solution is not there to reach some GHz goal, but rather to manage power draw, and fan size (as Mark Cerny explicitly said at 34:33).

He said that the wrong prediction of future power draw caused the fan noise, and that they needed to design unknown margins into the cooling.

They could have a 1000W constant power design and what he said still holds true.

 

[London Geezer]

Incidentally, "XBox Series X" is crap for internet searches! Search PS5 and you get PlayStation 5, but looking for the Xbox is unnecessarily awkward. "xb series x" seems the shortest that actually works?

Actually, this is Microsoft's fault. Their marketing department's job is to make sure that their Google search stays at the top - this isn't free of course. In my previous company, we had to do a huge amount of work to make sure that our name would be the firsr result when searching for it (it wasn't always the case), or that when searching our industry you would get our company as the first result - or at least the first three. Forget about the second page of Google search results, that's where companies die.

 

[Tkumpathenurple]

In all seriousness, if you use Bing for "Most Powerful Console" it brings up results for Xbox One X, no mention of Sony at all, and no mention of future consoles. Probably a bit of a bias in those results, though.

Must be. I searched for "PlayStation 1" and the only result that came up was text telling me the authorities have been informed.

 

[zupallinere]

Compounded by people also calling it SeX!
All because MS didn’t want to appear behind Sony with Xbox 2 (aka 360). Sad really, like you say, it makes finding anything Xbox related a PITA

From XBone to SeX it seems MS is trying it's hand at subliminal marketing.

 

[mrcorbo]

How many SEOFlops did they lose with their naming inefficiency?

 

[milk]

The naming convention is PS5's biggest advantage. Lol.

Yeah, but it still leaves sony in a performance diaadvantage since 5 is only half of X (10). And that's ignoring the fact that the new playstation only stays at 5 at peak clocks, but depending on workload it may throttle down to PS4.8, or even PS4.7! I'm waiting for more details fron devs on this, but I imagine it won't ever go as low as PS4.PRO. I mean, Cerny can't have fucked up this bad.

 

[Globalisateur]

How many SEOFlops did they lose with their naming inefficiency?

10% market share each generation. There is the color, too. studies have shown that the blue is the most visible, most impactful, color for the brain. Add 5% each generation.

 

[KeanuReeves]

He said that the wrong prediction of future power draw caused the fan noise, and that they needed to design unknown margins into the cooling.

They could have a 1000W constant power design and what he said still holds true.

And as he said, the problem then became that they had to use a bigger powersupply and fan just so they could reliably manage the extra power draw/heat.
So instead of doing that, they decided to go with lowering the clock when certain heat generating instructions are used too much.

 

[KeanuReeves]

We really don't know for absolutely certain which between cooling system consistency and performance goals was the chicken or the egg as a driver for the development of their solution. In the end though, it's both.

But there is no official statement to suggest that it was ever the frequency that was the goal. I think its more beneficial to take his 'actual' words on this instead of resorting to "it is implied that they did this for x reason".

The variable clocks is what allows them to reach the clocks they did with the cooling system and power delivery they implemented.

Because it makes powerdraw consistent in worst case scenarios, sure.

 

[Metal_Spirit]

Exactly, and that's what we don't know, how much lower the clocks can go. Some say 3Ghz for CPU and 2Ghz for the GPU. Which makes sense per github.

You keep comparing this with PC, and PC boost clocks. And it has absolutely nothing to do with that techology.
Read the Eurogamer article. They are quite clear on that.

 

[PSman1700]

You keep comparing this with PC, and PC boost clocks. And it has absolutely nothing to do with that techology.
Read the Eurogamer article. They are quite clear on that.

No im not comparing it to pc clocks. It's a fact the frequencies can go down on the PS5. It's basically boosting all the time, from what clocks i dont know, but some say its 2ghz (gpu) and 3ghz (gpu). When pushed too hard, clocks go down. By exactly how much isn't officially known, but the suggested 2ghz for gpu and 3ghz for cpu make sense.

 

[Metal_Spirit]

No im not comparing it to pc clocks. It's a fact the frequencies can go down on the PS5. It's basically boosting all the time, from what clocks i dont know, but some say its 2ghz (gpu) and 3ghz (gpu). When pushed too hard, clocks go down. By exactly how much isn't officially known, but the suggested 2ghz for gpu and 3ghz for cpu make sense.

First. Pushed too hard is what? According to Cerny that is when the worst case scenario used when creating the cooling system is passed!
So, it will hardly ever downclock.
Now... how much? You want something more official then the own Cerny’s words?
A couple of percentage points. That is enough to regain 10% power and put the system back on it’s budget.
As I say you are comparing this to PC. PS5 has no base clocks. PS5 like every other console is made to work at full power. Stating up to 10,28 Tflops is just a need because clocks are not locked. Otherwise Sony could be sued.

 

[PSman1700]

So, it will hardly ever downclock.

Then why even mention it? It will never downclock, but lets mention variable clocks.

A couple of percentage points.

If they could have sustained those clocks, they would, but they can't, so downclocking will happen when both the CPU/GPU are pushed too hard. Developers probably get to choose which of either they can fully utilise.

Stating up to 10,28 Tflops is just a need because clocks are not locked.

Exactly, because that 10.28TF for the GPU is not sustained, unlike for the XSX where they are (MS even advertised 'sustained' probably because they where up to date of Sony's variable clocks)

 

[itsmydamnation]

Then why even mention it? It will never downclock, but lets mention variable clocks.



If they could have sustained those clocks, they would, but they can't, so downclocking will happen when both the CPU/GPU are pushed too hard. Developers probably get to choose which of either they can fully utilise.



Exactly, because that 10.28TF for the GPU is not sustained, unlike for the XSX where they are (MS even advertised 'sustained' probably because they where up to date of Sony's variable clocks)

Whats pushed to hard? look at the ROME server SKU's if you dont use 256bit SIMD ( even full load 128bit SIMD) the 64core cpu's hold 3.2ghz at full load. move to avx256 and your down to around 2.4ghz clock.
Now look at what they said about miss projection for the cooling on PS4. Would you expect this generation to stay at 128bit ops or move to 256bit ops, that alone could easily account for all the down clocking requirements. I would assume it would stay 128bit ops but if it doesn't and everything moves to 256bit ops and it hasn't been accounted for my console just got broken.

Now remember the marketing side. in the ROME example AMD market that as a 2.x base clock, but intel would market that with a 3.2ghz base clock that has negative AVX offsets, but in effect the same behaviour.

 

[PSman1700]

Whats pushed to hard? look at the ROME server SKU's if you dont use 256bit SIMD ( even full load 128bit SIMD) the 64core cpu's hold 3.2ghz at full load. move to avx256 and your down to around 2.4ghz clock.

So what should the manual say, avoid using avx or else clocks will adjust? Even today there are games using it, my 3900x gets hotter but doesn't slow down (BFV).
There are probably other ways to max that CPU though, besides AVX instructions. And that's the CPU side of things.

Like i mentioned before, there is probably more to it then 'it basically never has to downclock, and when it does in such rare cases, its a couple of procent which equals 2%'
Alex explained it here before, i think that's very close to the truth.

 

[MrFox]

Nobody said it never downclocks. The quote to use is Cerny's. What's so difficult to understand?

This is hardware predicton. Maybe the average is 2% maybe it's more. I predict from this point in time all future comsoles will,use sony's method. It will become the most efficient use of silicon, power, and yield.