• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

Sony CEO: "We have not yet decided on PS6 launch timing"

So I'm going to install Linux on my PC and think it doesn't use the Windows driver?

Do you realize that the PS5's GPU is entirely controlled by power consumption?
And that a PC's GPU is controlled by temperature?
I'm not sure what you mean, that is Linux running on a PS5, no need to bring up discreet PC GPUs. And you're wrong anyways, most PC GPUs are power limited as well, not temperature limited.
The point is the findings there do not mean shit. It means absolutely nothing. Its not a "hey, look at how the PS5 APU really performs" argument but a "hey, look how Linux currently runs on the PS5 APU".

Unless you are running that thing exactly how the PS5 runs it using Sony's drivers, then it does not mean anything at all. If anything the fact that it runs as well as it does in what is no doubt an unoptimized mess speaks volumes.
The test is perfectly valid. It shows that the difference between 2Ghz and a dynamic boost is insignificant and that PS5 optimization on popular 3rd party titles doesn't really amount to much. Obviously so, considering it is basically just a PC.
 
It uses the mesa open source driver. And it runs identically to PS5 while at 2.0 GHz.

Now you can argue that the mesa driver is better than the Sony driver, but then that's arguing Console optimization magic is actually negative and PC runs faster.

You are also ignoring the fact PS5 Linux can't hold 3.5 | 2.23 GHz without thermal throttling. It's not simply that it runs identically to 2.0 GHz Linux PS5, but also that PS5 can't maintain 2.23+3.5 on Linux with maxed fans.

But again, you want to argue about this, shoot a DM. This isn't the thread for it.

Don't forget that secret spicy sauce inside the PS5 tho
 
Eventually RAM prices will come down because new manufacturers will come online but it might take years

PS6 in 2030 maybe
 
Cutting down on there specs due to ram price is silly.

Stick to your original vision, it will sell out the first year as the hardcore will buy it day one and like alot of consoles stock will be limited the first year.
 
Last edited:
The next generation is handheld PCs and streaming boxes.
ToFwcJZ1ZnMYRDIA.gif
EuyaCzhsKjO2msLf.gif
 
It uses the mesa open source driver. And it runs identically to PS5 while at 2.0 GHz.

bla bla bla

But again, you want to argue about this, shoot a DM. This isn't the thread for it.
Because you're going to send the Sony driver source code to prove that it's the same as the Mesa driver.
 
Last edited:
I'm not sure what you mean, that is Linux running on a PS5, no need to bring up discreet PC GPUs. And you're wrong anyways, most PC GPUs are power limited as well, not temperature limited.

The test is perfectly valid. It shows that the difference between 2Ghz and a dynamic boost is insignificant and that PS5 optimization on popular 3rd party titles doesn't really amount to much. Obviously so, considering it is basically just a PC.
I'm talking about the standard. I'm not talking about the guy who installs a water cooler that dissipates 5000w and then the GPU is limited by the power supply.

It doesn't mean anything, it just means how Linux runs on the PS5.

It's bizarre, as soon as Xbox went bankrupt and the Xbox fanboys disappeared, the PC fanboys started with the FUD about "RDNA 1.5," "fake clock," and "unsustainable performance". And you know what's bizarre? It has nothing to do with reality; you want to impose this narrative on the internet.
 
Last edited:
30GB of RAM was already a little disappointing, drop to 24GB would be bad, 20GB is joke territory.

This is supposed to be a next gen hardware, so I hope Sony stick with the plan.
 
It's bizarre, as soon as Xbox went bankrupt and the Xbox fanboys disappeared, the PC fanboys started with the FUD about "RDNA 1.5," "fake clock," and "unsustainable performance". And you know what's bizarre? It has nothing to do with reality; you want to impose this narrative on the internet.

Clocks based on power level and RDNA1 + RDNA2 RT architecture was always the reality.

Some people just wanted to not accept it...
 
What are you talking about bruh?

Remember the PS4? Remember it sounding like a jet engine? Ever wondered why that was?

Or you comparing 200W+ APUs to systems that pulled around 80W+ with the PS2?

Like you don't even realize that the fact the PS5 is using liquid metal as its TIM and is whisper quiet in an attempt to keep 200W system cool and run quietly is actually a remarkable feat of engineering, but you are here advocating for what? Thermal paste, extremely large heat sinks, and fans that sound like a hair blower?

Like what are you even talking about???

You misunderstood my comment, liquid metal was alright for the launch model so that Sony could ship something but now we are still stuck with this 220W+ APU.
For comparison the launch PS4 consumed 150W+ and they managed to shrink it to 80W so around 40% improvement, fast forward today and the PS5 Slim consumes around 10 Watts more than the launch PS5. I'm not bitching about liquid metal but its expensive and feels more like a workaround, but I'm glad that Sony is committed to keep the PS5 Silent and IMHO adds to it being a premium product.
 
The excitement level for a new generation has to be historically low. There are just not enough high quality games releasing. SONY studios themselves have been compkete dog shit with pumping and dumping live service games.

The ps6 will be the equivalent of upgrading ones graphics card and upgrading the Ram on PC.
 
If Sony will force downgrade just to force a 2027 launch, this might be the first time Sony will lose this gen. Its DOA. Just go with late 2028 or 2029 launch with just a few upgrades and modification but still be affordable.
 
The excitement level for a new generation has to be historically low. There are just not enough high quality games releasing. SONY studios themselves have been compkete dog shit with pumping and dumping live service games.

The ps6 will be the equivalent of upgrading ones graphics card and upgrading the Ram on PC.
To be honest, next gen will just be more frame rate, allot of AI designs, and path tracing, nothing else is new. If only VR was heavily supported by Sony, and developers, this could be the real game changer with so many untapped potentials. And sadly Im not excited with PS exclusives due to force and push for dei and woke especially with headquarters is still in leftist infested california. Hopefully changes from hoshino in the long run and back to solid and very based gaming of ps1 to ps3.
 
30GB of RAM was already a little disappointing, drop to 24GB would be bad, 20GB is joke territory.

This is supposed to be a next gen hardware, so I hope Sony stick with the plan.
So true. Vera Rubin has 288GBs. RTX 6000 PRO has 96GBs. 30GB is really disappointing.
 
You misunderstood my comment, liquid metal was alright for the launch model so that Sony could ship something but now we are still stuck with this 220W+ APU.
For comparison the launch PS4 consumed 150W+ and they managed to shrink it to 80W so around 40% improvement, fast forward today and the PS5 Slim consumes around 10 Watts more than the launch PS5. I'm not bitching about liquid metal but its expensive and feels more like a workaround, but I'm glad that Sony is committed to keep the PS5 Silent and IMHO adds to it being a premium product.
Ok... fair enough. But you are still wrong though.

First off, on the "PlayStation" side, your mistake is that you probably didn't know that the PS5 slim, was just a chassis revision; the actual APU/cooling/power revision already happened. The changes just stayed in the bigger chassis of the OG PS5 until the smil came along. The 7nm PS5 had a TDP of around 230W. The 6nm PS5 what was in the 1200 model and also subsequently in the slim, had a TDP of around 200W. So they could use a smaller cooling system, smaller chassis...etc. They likely just stuck with the liquid metal cause it wicks heat out quicker from the APU... its more performant.

And then secondly, and more importantly, the fab process. The fab process a chip is on determines how much power you need to get to any given clock. Like you see that PS4 you are talking about, the 150W down to 80W... that's in reality going from a 28nm fab process down to a 16nm fab process... vs in the case of the PS5, where they went from a 7nm process to a 6nm process.

And it gets even more complicated than that, one would think that as fab process shrink you can always just clock higher and draw less power, but its not that simple, as fab process get smaller and the physical chips gets smaller, pulling heat off the chips gets harder as the heat is concentrated over a much smaller surface area, a hotter chip means you need higher voltages to drive power through it, more power means even more heat....rinse and repeat.
 
The fab process a chip is on determines how much power you need to get to any given clock
Not just that. It depends on binning and physical design.

For binning, the key metric is parametric harvest %. Specifically, what % of known good dies are you planning to use? Since the voltage/frequency curve is calibrated for the worst die, you practically always have to disqualify at least the bottom 1%. But disqualifying 10%, 30%, etc. can yield a decent clockspeed boost at the cost of higher die cost.

Thus once you get your silicon back, what clockspeed you should run at becomes a purely economic decision.
 
Last edited:
I don't necessarily agree. I think the reality is that the PS6 will be positioned as an enthusiast device. You want to play the games but don't want to spend $800 for it? They have a console for you, it's the PS5 or PS5 Pro. Those will get all the same games.
What? No chance, is the PS5 enthusiast level? PS4? Even 3? 2 maybe? What about 1?

No, they are mass consumer level devices, I see no reason Sony would change that strategy, a strategy that has done them extremely well for 5 console generations (3 is about as close as you could get to a more enthusiast level and guess what? It was the lowest selling, lowest profit (heck, they lost money) generation of all)

The Pros are the enthusiast level, devices they can release mid gen for a profit and even then don't sell as well as the base console, but that's ok.... Because they sell them for a profit.
 
And then secondly, and more importantly, the fab process. The fab process a chip is on determines how much power you need to get to any given clock. Like you see that PS4 you are talking about, the 150W down to 80W... that's in reality going from a 28nm fab process down to a 16nm fab process... vs in the case of the PS5, where they went from a 7nm process to a 6nm process.

And it gets even more complicated than that, one would think that as fab process shrink you can always just clock higher and draw less power, but its not that simple, as fab process get smaller and the physical chips gets smaller, pulling heat off the chips gets harder as the heat is concentrated over a much smaller surface area, a hotter chip means you need higher voltages to drive power through it, more power means even more heat....rinse and repeat.

Right but so far Sony seems to not be able to get their hands on anything below N6 from TMSC for their mass produced PS5 line (expect PS5 Pro) even though mid range Mobile SoCs are already using N4 so what is going on with Sony's supply side?

A PS5 Slim with N3 process would be amazing and potentially use below 100W power.
One way or another, AI and Mobile suck up all the N3 capacity from TSMC and I think Sony is having difficulty getting in there, so the PS6 will have to wait some more years.
 
Right but so far Sony seems to not be able to get their hands on anything below N6 from TMSC for their mass produced PS5 line (expect PS5 Pro) even though mid range Mobile SoCs are already using N4 so what is going on with Sony's supply side?

A PS5 Slim with N3 process would be amazing and potentially use below 100W power.
One way or another, AI and Mobile suck up all the N3 capacity from TSMC and I think Sony is having difficulty getting in there, so the PS6 will have to wait some more years.
Oh sony can get their hands on it just fine. There is a catch though. Here is the per wafer cost of your fab nodes

  • 28nm ~$3k (PS4)
  • 16nm ~$4k (PS4 slim/PS4pro)
  • 7nm ~$10k (PS5)
  • 6nm ~$12k (PS5 slim/PS5pro)
  • 4/5nm ~$20k
  • 3nm ~$25k (PS6)
  • 2nm ~ $32K
  • 1.8nm ~$45k
So the problem isn't that Sony can't get their hands on chips from the smaller fab nodes, it's that the costs are prohibitively high. Especially when you consider that even though the nodes are shrinking, because of the parts of the chip that don't shrink as well with each process node (basically the non-logic parts) we aren't seeing equivalent chip side reductions in relation to node reductions.

So once upon a time,you could be spedning 25% more on your wafer with a node shrink but be getting 80% more chips from the same wafer. But now, you could be spending as much as 110% more on a wafer to get around 60-80% increase in production yield. Not a lot of incentive to bother, is there?
 
Right but so far Sony seems to not be able to get their hands on anything below N6 from TMSC for their mass produced PS5 line (expect PS5 Pro) even though mid range Mobile SoCs are already using N4 so what is going on with Sony's supply side?

A PS5 Slim with N3 process would be amazing and potentially use below 100W power.
One way or another, AI and Mobile suck up all the N3 capacity from TSMC and I think Sony is having difficulty getting in there, so the PS6 will have to wait some more years.
Die shrinks don't make sense anymore because cost per transistor doesn't go down.
  • 28nm ~$3k (PS4)
  • 16nm ~$4k (PS4 slim/PS4pro)
  • 7nm ~$10k (PS5)
  • 6nm ~$12k (PS5 slim/PS5pro)
  • 4/5nm ~$20k
  • 3nm ~$25k (PS6)
  • 2nm ~ $32K
  • 1.8nm ~$45k
Idk what's your source but it's not *that* bad
 
Loose to whom?

There is no competition
Pc. And pc proxy.
Sony's bread and butter is too costly aka AAA sp games. If the I industry is ps/pc for release then it's a growing problem due to open (zero paid online) vs closed (no nsfw mods/psn).

When affordability becomes an issue folks will look for alternatives.
 
Die shrinks don't make sense anymore because cost per transistor doesn't go down.

Idk what's your source but it's not *that* bad
This is one of my sources... and while the prices here show the lowest ranges, I have other sources that quoted higher... so I listed the worst case scenario prices I found.
 
What? No chance, is the PS5 enthusiast level? PS4? Even 3? 2 maybe? What about 1?

No, they are mass consumer level devices, I see no reason Sony would change that strategy, a strategy that has done them extremely well for 5 console generations (3 is about as close as you could get to a more enthusiast level and guess what? It was the lowest selling, lowest profit (heck, they lost money) generation of all)

The Pros are the enthusiast level, devices they can release mid gen for a profit and even then don't sell as well as the base console, but that's ok.... Because they sell them for a profit.
We'll see. Gonna be hard to sell the mass consumer on a $800 PS6 when NBA2K runs great on a PS5.
 
You misunderstood my comment, liquid metal was alright for the launch model so that Sony could ship something but now we are still stuck with this 220W+ APU.
For comparison the launch PS4 consumed 150W+ and they managed to shrink it to 80W so around 40% improvement, fast forward today and the PS5 Slim consumes around 10 Watts more than the launch PS5. I'm not bitching about liquid metal but its expensive and feels more like a workaround, but I'm glad that Sony is committed to keep the PS5 Silent and IMHO adds to it being a premium product.
It doesn't make sense anymore to keep shrinking the APU when the investment to do that is not worth it anymore. PS5 APU will remain at 6nm.

It's not worth it because any savings you gain from it will be eaten away by RAM cost, let alone the hundreds of millions cost for the R&D. It's not like before where shrinking the APU chip will get you $299 console because the main chip used to be the most expensive part that needs cost reduction.
 
Question though. How do you suppose that extra RAM will be put to good use with LLM when PS5 can't handle it and cross-gen is here to stay for long. You don't reckon devs won't bother because their games have to run on PS5 too? Unless they decide it's okay to have two different games and experience.

I guess if LLM is truly in the cards for next-gen then PS6 game will truly feel different from its PS5 counter part, on top of realistic lighting with path tracing. PS5 version of games will feel not only inferior graphically, but almost feel like a different game.
They wont be supporting the PS5 for the full generation, maybe 4-5/8 years and then they need the system to run cross gen games in the PS7 gen(so you're looking at a system thats going to last at least 10 years). So you cant bet it all on 20GB of memory not even 24GB and expect it to last 10 years, they'd need something like 30-36GB of memory which cost can be amortized over the 10 year life of the console. And the games can be designed around the base PS5 during the cross gen period then upscaled to higher performance/resolution with the PS5 pro and PS6 while leveraging better RT on the PS6, higher quality textures due to higher memory and acceleration for neural based rendering techniques on the PS6 if possible. More memory makes all this more feasible.

Its not just LLMs but all kinds of neural networks and all kinds of workloads. There are JEPA neural networks which could be very useful for gaming since they understand real world physics better and are better at generating images and videos efficiently than transformers for example. JEPA models are more compute efficient for video and image than Transformers but still for denser larger worlds require more memory there's no way of running away from that.

Sony is at a difficult point because if they dont ship with enough memory they will have a console with the compute but not worth upgrading to since consumers wont see a meaningful difference between the PS5 and PS6. A huge memory jump can greatly help with that. But in the short term memory has gotten very expensive and is likely to continue like this all the way into 2030. So its not an easy situation for them tbh. Because once you ship with 20GB of memory you're stuck with that system for about 10 years.
 
PS5 exposed with 2ghz (9.2TF FP32)?

75b5d3d54441e9cdc61bdcf2c73119304fde373793b853a489389c4ad9ba6525.gif
Well I always thought this was the case, but in real world performance it has been running games just as good as the Series X, MS paid for extra silicon then hamstrung it with the Series S. So really PS5 has been the benchmark for stable current gen gaming. And PS6 will likely be the same.
 
Well I always thought this was the case, but in real world performance it has been running games just as good as the Series X,

How games run on these boxes (relative to each other) is mostly about optimization on the developer side and very little to do with specs.

Same on PC, some games run better on AMD not because the GPU is better than the Nvidia equivalent, but because the game code is better suited for that part.
 
We'll see. Gonna be hard to sell the mass consumer on a $800 PS6 when NBA2K runs great on a PS5.
That's why PS6 sales will be nothing to write home about. The technological jump will be to small for the mass market to care and the costs will be too high. I also don't really see any must have software landing anywhere close to launch. Everything third party will be cross gen.
 
Last edited:
since consumers wont see a meaningful difference between the PS5 and PS6. A huge memory jump can greatly help with that.
Okay. What other real world perceivable difference would a 30GB PS6 have versus the PS5 that a 20GB PS6 won't have. 20GB PS6 is perfectly fine for gaming path tracing.

I was sold on LLM when it was mentioned. But due to cross-gen with PS5, you won't
see that anytime soon I suppose.

And how about the 24GB PS6 handheld. With this LLM and other ML innovations requiring 10GB and more, how is it able to keep up.
 
Top Bottom