Intel is shooting for PlayStation 4, but has it scored?
A new rumor suggests that Intel has succeeded in its efforts to place
its forthcoming discrete GPU product, codenamed Larrabee, in the next
generation of Sony's PlayStation game console.
By Jon Stokes | Last updated February 6, 2009 2:20[/b]
I've known for a while, and even mentioned on Ars, that Intel has a
team that's actively courting the Sony PlayStation 4 contract, and
Intel VP Pat Gelsinger affirmed to me in an interview that the company
would love to be in a next-generation console. But The Inquirer's
Charlie Demerjian has published a whale of a post-CES rumor claiming
that Intel's forthcoming Larrabee GPU did indeed score the GPU spot on
Sony's next-gen console.
There will be a ton of skepticism over this, and for good reason=97I
myself have actually had informal, off-the-record conversations that
indicate to me that there may not be a lot of love for Intel on the
technical front inside of Sony. But this rumor has a lot going for it.
I'll enumerate the plusses and minuses below, but before jumping in
let's hand the mic over to official Sony PR for their take.
"We do not comment on rumors and speculations. Our utmost priority is
to further increase the installed base for PS3 and PSP, with more new
and exciting entertainment and services (game/non-game and package/
network)," a Sony spokesperson told Ars. "PS3 is still in its third
year from launch and it is too premature to talk about the next-
generation home platform. At this time, nothing concrete has been
I actually believe that nothing concrete has been decided. But Sony
may be leaning one way or the other. Let's look at the forces at work.
It's all about the (rapidly shrinking) bottom line
The number one, overriding factor in any decision that Sony makes
right now is money=97and I don't mean, "how will we grow our console
marketshare and conquer the living room," but "how will we stop
hemorrhaging money and ensure that, whatever happens with the next
generation of consoles, we attain profitability with the PS4 as
quickly as possible." Given this new reality, I find Charlie's claim
that Intel essentially bought its way into the PS4=97technical
considerations be damned=97very convincing.
This "ten-year console" stuff, where Sony looks at a console's overall
profitability over the course of a decade? That's the kind of thing
you can afford to do when you're not staring into the abyss. Intel has
deep enough pockets and enough fab capacity to take a real hit to its
margins on Larrabee and offer Sony a cheap GPU with guaranteed
availability. If any semi maker is going to survive this recession and
come out the other side still fully invested in the success of a
discrete graphics product, it will be Intel.
As for the Kuturagi-style grand visions of Pixar-quality graphics in
the living room, and networked consoles all changing the way we
entertain ourselves, that, too, is bull-market bull. With the PS4,
Sony will just want to stay in the game, and have a very quick path to
profitability. One might even go so far as to say that, at this point,
Sony execs probably wish they had gone the Wii route.
Note that these same financial pressures also drastically change the
cross-platform development equation from the console vendor's
perspective. In the distant, more profit-friendly past past of early
2008, console makers still courted exclusive titles, and Sony in
particular counted it as a major plus that games optimized to take
full advantage of PS3 wouldn't be easily portable to Xbox or the PC.
But in today's world of shrinking margins, Sony's incentives have
shifted 180 degrees, and with Larrabee (especially if it's paired with
a Nehalem-derived CPU), Sony can actually boast, "write it once, run
it on the PlayStation 4 and the PC," to developers.
To sum up, Sony has paid a heavy price for its efforts to attract
exclusive titles to the PS3 by promising developers superior visuals,
and that's a price that the company cannot afford to pay with its next
What about Cell?
My guess is that if Sony were free to fantasize about the console that
it would ideally like to produce with PS4, it would probably just
double down on Cell. And why not? The same things that make Cell hard
to program will also plague Larrabee. The programming problems that
come with the many-core paradigm=97a paradigm that both Cell and
Larrabee fit under=97are fundamental, algorithm-level design problems
that are essentially ISA-indifferent. (Yes, this is a shift from my
previous "x86 will own the GPU stance.")
So Sony could safely just do more of Cell with PS4 by just increasing
the number of dedicated vector units and general-purpose cores. This
would also give the company free backwards compatibility with existing
PS3 titles, and, of course, enable it to leverage existing investments
in the PS3 developer toolchain and in Cell itself.
Ultimately, the question of which horse Sony bets on next hinges on
the numbers, and specifically on whether Intel is able to paint
Larrabee as the cheaper option going forward, even when the cost of
ditching the hardware and software investment in Cell is factored in.