PS4 to be based on Cell?

If SKUs don't come with local storage included, DLC won't sell nearly as well.

 

Not quite. All those things were important, but far more important as an idea than the actual execution. Marketing had a much bigger part than R&D in making the Wii the new 'fun' machine. I'm not saying that Sony won't need some catch to hit a similar market, I just don't think it'll come from R&D.

 

Who knows, they might repackage GC or at least its GPU into their next gen handheld or something. They are good at marketing.

 

given what we know in the pc.watch.impress.co.jp article, a PS4 equipped with a 16-SPU Cell (256 KiB per unit), 2 GiB of RAM (fully and openly shared with an nVidia GPU) and a 256 GiB SSD due in 2012 sounds about right ... all for $299!

 

I think they should either dump Cell and go for a more friendly commodity CPU + GPU combo. Or leverage SPU ISA and develop GPU, much like how Intel is doing with x86 and Larrabee. Even if it breaks backward compatibility with PS3. And they should only develop a Cell based GPU if it offers them cost advantage.

Sony can’t embrace both the way they did with PS3. GPU will become more and more programmable and will make Cell redundant and another source of headache. It offers no advantage to Sony or Developers.

 

What went wrong was, even though PS3 had a powerful CPU with CELL and high capacity Blu-ray, the rest of the architecture, GPU rendering power, system bandwidth, graphics bandwidth, etc was not *as* massive a leap beyond PS2, that PS2 was beyond PS1.

 

As long as the PS4 is like the PSX to DC then I will be happy. I would be way happier if it were like PSX to PS2, but if I can't have that then I at least want something.

 

Indeed, CELL was spec'd for at least 4 GHz, maybe as high as 4.6 GHz.

I remember reading that IBM tested SPEs at 4.6 GHz.

Then at E3 2005, Kutaragi announced CELL would run at 3.2 GHz for computer entertainment applications.

I'd be shocked if PS4 CELL wasn't 4 or 5 GHz.

 

But in fairness that would at least be a decent upgrade to the DS, so thats fine.

 

I dont get why alot of people are assuming there will be an SSD drives put in consoles this time round. Why would sony go to all the trouble of making an effiecent cost effective platform and more then blow the rest of the budget on an SSD? SSds will be great in the next next gen 2017.

The funny think thing about SSDs anyway is that if you get a really really cheap SSD drives (think newSupertalent Drives just released) they are still way more expensive than normal drives and aren't even better performing.

http://www.dailytech.com/article.aspx?newsid=13087

I think Sony should just carry on what they are doing. Put a cheap standard drive in, And let us upgrade ourselves to a 2TB drive (Nicely done for 2011)

 

My point is that Blu Ray was the reason that the things you mentioned were compromised to achieve cost reductions

 

I don't think for a moment that Sony will rush the PS4 and abandon their track record of making very powerful systems. The PS3 is a powerful system (especially the CELL) and it has many more years of life in it. I think they will make various slim versions of the PS3 until about 2014 when they launch the PS4.

I think the PS4 will be incredibly powerful with a monster CELL chip or multiple smaller cell chips.

 

What make you think they will marry it with a contemporary GPU design?
The Wii didn´t use a contemporary GPU and it worked quite well.

Yeah once the developers have got used to the restrictions of OOOE it doesn´t make sense to go back.

As IBM already has implemented a DDR2 interface in their HPC version of Cell it´s probably does not involve much work reshaping it for DDR3. They may even go with IBMs improved DP SPEs.

Thanks to one for the translation, clarifying which parts that are Gotos own speculations. The article does not really provide much more information than Sony is "evaluating different solutions for the PS4".

However, I agree with one part of Gotos speculation: "The only option that makes the 2011 launch possible is to refine the current architecture". I think that is true, if a 2011 launch is the target that is.

BTW I am backing off from my previous bold statement that such a solution may not be called PS4 but "PS3+something". It may be called PS4 for marketing reasons, but I still see a possibility that it will be distributed on the same BD disk as the PS3 game. As a lot of graphics and sound assets on the disk can be shared between the two versions, the game will hardly be limited by the 50 GB of a two layer BD disk

I think that would make the path to upgrade from PS3 to PS4 much simpler. From one point in time you start to get a PS4 game bundled with every PS3 game you buy and once you buy a PS4 you can replay the game and have more refined graphics. IMHO that would be quite an incentive to upgrade. .

 

I find the discussion about a "Wii"ified PS3 pretty interesting and could not that bad perf wize as some make it sounds.
Some part of the article are pretty much irrelevant like the X2 statment... it means basically nothing and should not be discussed as it will go nowhere.

I think that the main issue for the system could be if Sony aims at full hardware compatibilty.
That would imply that the GPU would be an evolution of the RSX.
That could really be the limitating factor for the system performances.
If sony don't care about it or consider providing Backward compatibility for limitated number of titles well they have quiet some possibilities to offer something both good enough and cheap.

In fact depending on their budget R&D and silicon, it could be cleverer than a lot here seem to think for Sony to keep an "almost actual" cell as CPU in their next system.
A lot of persons here expect Ms to go with a "GPU heavy" design, Sony may well come with the same conclusion (no matter there silicon budget).
I somewhat disagree with Nao, ie "a cell stuck to 2010/11 GPU doesn't make sense".
Ok stick a huge cell to a huge GPU might not be the cleverest thing to do but but how about the cell we know?
The actual might well up be to the task to be the "brain" of a GPU heavy design.
It will be pretty well know by the developers by the time Sony launch its next system.
It will benefit without a change (ouvious ain't it ... ) from all the existing tools.
It has the power to do so!
And last but not leats it would be a tiny/cheap/cold chip! freeing some budget in what could really well be a mid to low budget system.

 

Because I just don't see them re-using RSX, but yes, they might even try to use it again. On the other hand Sony is not Nintendo, they can try to adhere to Nintendo philosophy in order to save costs, but they can't execute on it as they were Nintendo.
People need a very good reason to buy a new underpowered Sony console imho, and that can't be Mario or Zelda.

I'd like to meet these developers Perhaps you wanted to write "restrictions of IOE"..

 

nAo, aren't you going to pull some strings to get your assessment into that alleged PS4 developer feedback form ?

 

I wish I was that much influent I'm just speculating here, I don't really know what Sony is doing right now.

 

nAo, when you mention that Cell+ doesn't make much sense when connected to a future GPU, I assume it's because of CUDA and similar technology that seems to overlap with Cell's functionality.

What would be, in your opinion, the way to go for Sony, assuming they want to maximize performance (while keeping below this gen's launch price points, perhaps $400 maximum)?

Could they use a custom GPU, dedicated exclusively to graphics processing and possibly simpler than the monster GPUs we'll surely have around 2011, keeping a more powerful Cell for the rest? Or would dumping the Cell and using a multi-core general-purpose CPU coupled with such a monster GPU be better?

 

Yeah I know, but a real and sincere feedback is fair game anytime. You probably have more access to Sony than most of us here. ;-P

Just a wild thought.