By using this site, you agree to our Privacy Policy and our Terms of Use. Close

Forums - Sony Discussion - Sony Moving Away from Cell-based PS4

Source: www.andriasang.com

 

Technology writer Hiroshige Goto has shared a few insider tips on the future of the PlayStation brand in the latest installment of his Weekly Overseas News column at Impress Watch.

Previously, Goto reported that Sony was looking both into a Cell-based and Intel Larrabee-based architecture for PlayStation 4.

Regarding Larrabee, in summer 2008, Sony began looking at the architecture, apparently because high ranking officials at Sony had expressed interest. Now, it appears that Larrabee is out of contention due to 3D graphics pipeline performance issues and low power efficiency.

Regarding Cell, some time in 2008, the company asked game publishers what they'd want of a Cell-based PS4 in terms of number of SPU co-processors and what kind of programming difficulties they've been having with the current design.

According to Goto, Sony, IBM and Toshiba recognized some of the problems programmers were having with Cell and, early on, came up with a couple of plans for fixing the issues. Included in these was a plan for something that was at one point called "SPU2." This new version of the SPU would shift the 256 kilobyte local store space that's included on each SPU chip into the role of a hardware management cache, allowing the SPUs direct access to main memory and allowing programmers to program for a single memory space, similar to a standard PC CPU (for Japanese readers, the original article explains this in much greater detail).

Goto says there are signs that some time this summer, Sony was looking into using this updated SPU design in the core of the PS4. At the very least, he says, the design was a strong candidate.

However, he's recently started hearing about other plans. While he's unable to get into the specifics, at a broad level, the new plans call for a PC-like multicore setup.

Goto also provided a few guesses as to when we'll see the next generation of hardwaree. He believes that because Sony, Microsoft and Nintendo, are looking into architecture solutions now in 2009, the next generation will come around 2012. The reason for this is that it takes 24 months to take a system from concept to production.

In addition, he expects next generation portable systems like PSP2 to precede the release of next generation consoles.



Around the Network

Firstly, while Goto is a good journalist, he doesn't usually have secret industry sources. He goes on very good guesswork and public domain information.

There are only four real choices of CPU for a performance-oriented PS3.

- Intel
- AMD
- IBM (Standard POWER)
- IBM (Cell)

Any other CPU vendor does not have the performance to compete. I'm thinking of ARM and VIA here primarily. Those three companies also have the strongest CPU manufacturing capability; choosing another company would mean negotiating with one of them for fab space anyway, because independent foundries like TSMC or UMC are too low-power oriented to do powerful CPUs in a reasonable time and budget.

So, the top two have the advantage of being x86 and hence easy sharing with the PC platform and easy to develop for as most developers understand x86 multi-core. The Cell would also be easy to program now, but since they would have to make architectural changes like the cache it would mean another learning curve, which a third-place entrant into next generation cannot afford to impose. And standard POWER would mean easy sharing with Xbox 360, though I believe MS will pick an x86 CPU this time around since it's their native platform and they only went away from it because Intel's architecture in 2004-5 when they were designing wasn't good enough.

The advantage of AMD (though for political reasons I think it's the least likely option) is that you get a great-performing GPU and CPU on one die from one foundry. That will bring down dev costs and manufacturing costs considerably; all previous Playstations trend towards greater integration like that. Although Intel can too, their GPU isn't good enough and the integration is on the package level rather than the die.

There are only two choices of GPU, now that Intel's Larrabee has been delayed over and over and indications are that its first iteration was hot and underperforming. Not something to risk a console on; at least AMD's Fusion uses a known CPU and known GPU, just new process and die layout.

- AMD
- Nvidia

Sony's choice of Nvidia was wrong this gen; the PS3 GPU is inferior to the AMD one. Nvidia's roadmap is also not strong, with the GT200 having been delayed 7 months and Fermi being similarly delayed and not out - and with a huge TDP of 225W it doesn't sound like derivative parts will be as low power consumption as is required for consoles. So I would say AMD is the most likely option. There's also the chance of a deal with AMD for both CPU and GPU too.



best move ever cell sucks for game programming the industry is not rdy for it



4lc0h0l said:
best move ever cell sucks for game programming the industry is not rdy for it

Uncharted2 craps all over your opinion.

Plenty are ready for it and have taken advantage of it. 



Sucks for developers who put their time and effort into learning how to program for the cell.
The games were starting to look really nice and gave Sony a kind of exclusivity that the other consoles didn't have.



Around the Network

Seems like a waste of money, they should just slightly upgrade the cell and save a ton of money.

By the time this gen is over Developers will be use to the cell, just like how the PS2 was, backwhen it first came out Developers were actually saying the PS2 was harder to make games on then the PS3 was. So they'll get use to the cell.



The cell is the main thing which makes PS3 stand out, if PS4 does not use it, it will not stand out, Sony have to think carefully before making any decsions with M$ around.



it's the future of handheld

PS VITA = LIFE

The official Vita thread http://gamrconnect.vgchartz.com/thread.php?id=130023&page=1

Hus said:
4lc0h0l said:
best move ever cell sucks for game programming the industry is not rdy for it

Uncharted2 craps all over your opinion.

Plenty are ready for it and have taken advantage of it. 


lol sony put so much effort on the cell if they had done the same effort on a powerfull video card they could have done the same or more than uncharted 2 with less effort .... and naughty dog has A LOT more budget for developing games than normal studios... so no the industry is not rdy for a processor like the cell and that is shown with crappy ports that still happen today, because developers still use the main procesor of the ps3 and not the spu... and 1 game doesnt mean anything when u got an industry that is 100 times bigger than just uncharted 2

hey and for your information IBM OFFICIALY dropped Cell arquitecture =) because they realise it wasnt the way to go

http://www.hardmac.com/news/2009/11/24/ibm-to-drop-the-cell



4lc0h0l said:
Hus said:
4lc0h0l said:
best move ever cell sucks for game programming the industry is not rdy for it

Uncharted2 craps all over your opinion.

Plenty are ready for it and have taken advantage of it. 


lol sony put so much effort on the cell if they had done the same effort on a powerfull video card they could have done the same or more than uncharted 2 with less effort .... and naughty dog has A LOT more budget for developing games than normal studios... so no the industry is not rdy for a processor like the cell and that is shown with crappy ports that still happen today, because developers still use the main procesor of the ps3 and not the spu... and 1 game doesnt mean anything when u got an industry that is 100 times bigger than just uncharted 2

hey and for your information IBM OFFICIALY dropped Cell arquitecture =) because they realise it wasnt the way to go

probably because they didn't have a major buyer for the chips, AKA sony.



I don't understand much about it. But, going for a x86 plataform would make the system cheaper to make? Since most of the processor produced nowadays uses this architechture. And using a x86 the development costs and time would drop? Since we see powerhouse PC games that costed little to make when compared to powerhouse console games.