Linked by Kris Shaffer on Thu 24th Mar 2005 19:07 UTC
Editorial Wishful thinking? Yes, but let's consider the possibilities. The last couple years have seen significant advances in hardware production and design. One of the more interesting (and potentially revolutionary) developments to take place this past year is the announcement of a new CPU, the STI (Sony, Toshiba, IBM) Cell processor.
Permalink for comment
To read all comments associated with this story, please click here.
Re: pointless speculation and hand waving
by KaS_m on Fri 25th Mar 2005 01:57 UTC

Man, am I sick of seeing comparisons between EE and Cell. The two situations are NOT comparable, people! There are three major differences between Cell and the "Emotion Engine".
A. Cell technology WILL deliver superior performance
B. Cell is absolutely guaranteed to be cheap and ubiquitous.
C. EE was hype, Cell is Sony's new long term strategy.

Cell's technology, as revealed by its patent, is much more impressive than EE was for its time. EE is a MIPS processor with extensions to its instruction set. Admittedly, it is set up in a very strange configuration in the PS2, but strange is very different from obviously superior. Consider the fact that Cell will ship with a POWER4 or POWER5 core, possibly dual-core, clocked at over 4GHZ in 2006. If that was the entire thing, it would be respectable, certainly the greatest POWER chip so far. It would be competitive with the offerings of AMD and Intel at that time. Since Cell is also going to come with 8 SPU vector processors, even if those SPUs are assumed to be completely idle when there's no OpenGL window onscreen Cell would still be great. While Cell would be competitive with Intel and AMD for general computing tasks, it would still absolutely destroy them in gaming performance, and therefore x86 would lose the entire enthusiast market (after the requisite inertia). Since the vector processors, in reality, will be useful in a very wide range of other tasks (pretty much all things that really require a fast CPU these days, actually), there will be no competition as far as performance is concerned. How can you compare this to EE, which was never even aimed at the general computing market?

Cell claims a good number of transistors, but not any significant increase over current CPUs. It comes in a sensible package and has a sensible thermal profile. Therefore, it is safe to assume, with technology posing no obstacle, that these will become very competitively priced very fast. The reason is obvious: Sony is sticking this thing in every HDTV, PVR or other media device starting 2006. It will be manufactured in volume unprecedented for general-purpose CPUs. It will be manufactured by IBM, and any other reputable chipmaker that wants to get on board. PS2's EE was needed for a much tinier market.

One final thing distinguishes Cell from EE most of all: the sheer scope of the project. EE was developed for a single division of Sony. Cell is depended on by Sony's entire high-end consumer electronics business, which is more than 10 times the size of its games division, and Toshiba's consumer electronics too, and whatever IBM sees fit to use it for, and for the countless other companies that will want to jump on the bandwagon immediately through licensing. Sony's entire future as a company rests on this. Most people assume the Intel/Microsoft duopoly is invincible, but here we have a company that is far larger than either that will be throwing all its immense resources at the project, and will have money pouring in from its great success in the consumer electronics space that they can use to support it throughout the initial resistance they will inevitably face in the general computing market.

That said, I don't want to sound stupidly optimistic. Unless Microsoft immediately gets behind Cell with a port of Windows, Cell will see very lackluster adoption for general purpose desktop computers. Linux is terrific, but it's adoption on the desktop is glacially paced and will continue to be so no matter what hardware comes out.

Nevertheless, Cell will be HUGE in high-performance computing. It will be HUGE in consumer electronics, and HUGE in gaming (Xbox Next's technology is so conventional by comparison, and its foothold in the market so tenuous, it can't possibly compete, Halo 3 or not). Comparing this new technology, revolutionary for vectorizable applications, to a MIPS knockoff is as ridiculous as comparing this new market behemoth to Sony's Playstation Divison. In case you noticed, the only people shoveling EE hype were Sony's Playstation Division, and the only people swallowing it were PS2 fanboys. The mere fact that this is being discussed on OSNews (and CNet and PC Magazine and everywhere else for that matter) should indicate to you the fallacy of the comparison.