View Full Version : IBM, Sony, etc.'s Cell Chip...

02-08-2005, 01:18 PM
Any comments on this new technology?

Some are claiming this will take out Win$Tel.

We tend to agree.

Note that APPLE and IBM are in cahoots and AMD is involved too.

Cell Chip technology literally changes everything. It is similar to how every cell in our bodies carries full instructions for a whole new us AND every cell has everywhere included-middle network ('self-organizing') associativity with every other cell (up to globally) in quasi real-time.

Imagine a MAC Mini with 100-1000 of these puppies in it!!!

What do you think?


02-08-2005, 03:25 PM
I think there are a dozen potential technologies like this, and few if any will ever actually become usable.

Craig R. Arko
02-08-2005, 04:03 PM
I think I'll defer judgment until some more technical whitepapers from IBM, Sony, and Toshiba are published.

I think this will be an important chip but a little more hard data and a little less marketing hyperbole is called for. :)

It doesn't take a huge leap to guess that things like Core Image, Core Video, and Core Audio could make use of the DSP-like capabilities of the Cell. How much use will depend on the aforementioned hard data.

Here are two looks at the Cell:



But neither can be considered authoritative.

02-08-2005, 04:46 PM
Yeah, but,

What's really omnifferent here?

We've BEEN classically modeling objects. (And IBM just dumped their PC business.)

Now, IBM, et al., are mod-alling VALUE, where Value is not objective properties, rather INTERRELATIONSHIPS!

This is a killer for all OO and classical stuff, period. Say goodbye to formal systems! This does for EVERY app what www did for Earth! (Yep, there is a tad o hyperbole here...)

Just remember this thread five years from now...

LoL :eek: ,


Craig R. Arko
02-12-2005, 08:52 AM
A good introductory look (http://www.realworldtech.com/page.cfm?ArticleID=RWT021005084318) at the CELL architecture.

02-12-2005, 03:49 PM
Hope Wang doesn't mind a quote here,

"As described previously, the CELL processor with 8 SPE’s operating at 4 GHz has a peak throughput rate of over 256 GFlops. To provide the proper balance between processing power and data bandwidth, an enormously capable system interconnects and memory system interface is required for the CELL processor. For that task, the CELL processor was designed as a Rambus Sandwich, with Redwood Rambus Asic Cell (RRAC) acting as the system interface on one end of the CELL processor, and the XDR (formerly Yellowstone) high bandwidth DRAM memory system interface on the other end of the CELL processor. Finally, the CELL processor has 2954 C4 contacts to the 3-2-3 organic package, and the BGA package is 42.5 mm by 42.5 mm in size. The BGA package contains 1236 contacts, 506 of which are signal interconnects and the remainder are devoted to power and ground interconnects."

How many times have you seen interfaces to ORGANICs before? Assuming Wang intends organic in a biological (bio(non)logical) sense.

Craig's link is excellent. Also, there is a lot of good material at EETimes.

See if you find any similarities to work we did back in 2000 here:


Look at our Self Organizing Network graphic about 2/3s down page. ;)

Thanks Craig!


02-12-2005, 04:16 PM
How many times have you seen interfaces to ORGANICs before? Assuming Wang intends organic in a biological (bio(non)logical) sense.
Well, the word is used in the sense of organic chemistry - i.e. materials derived from carbon compounds. An "organic package" as opposed to a "ceramic package". This is nothing particularly new - Intel has been using organic (resin) packages for a long time. See this Intel doc on the evolution of microprocessor packaging:

02-12-2005, 07:17 PM
From Robert Cringeley's recent column:

"Now to the Cell Processor, which looks to be the poster child for the technical revolution that's beginning. The Cell is great, the Cell is powerful, the Cell is scalable, the Cell burns 130 watts. Don't look for a truly mobile Cell ANYTHING for at least three years.

And people who think because it will first appear in a video game means that the Cell is inexpensive are wrong. Two hundred twenty-three million transistors cost the same to produce no matter what they are being used for. Sony will just bury the cost of the Cell and try to make it up through game royalties."


02-13-2005, 03:50 PM

We wondered about that. Haven't spent much time on hardware and firmware recently, but this 'cell chip' turned us back on. We were thinking rat neurons and digital hybrids. Sort of like Affymetrix on steroids. :p


We read Cringely's article. He is just superb, in our opinion.

Our hermeneutic of organic in any sense of a 'cell' chip carries extended semantics. One recent one is University of Florida's work hybridizing digital with neurons from a rat's brain. They apparently used it to successfully control an F-22 simulator. That impressed us, and it asks us to Cringelyesque ponder what this 'cell' chip is really all about. It looks like a neuron to us!!!

You may have read Jeffrey Satinover's The Quantum Brain. That's essence of which we speak. Part of that stimulated our everywhere association text in our graphic above. Architecture of these novel 'cell' chips appears implicitly 'everywhere associative,' a prerequisite for AI (e.g., Honda's robot glorified...) and formal emulation of quantumesque brain architectures (emerscitectures, i.e.). (Plus we're dreaming of our quantumesque MAC GN googolie. :rolleyes: )

10 years ago we were in a Director's slot on a system team for a Fortune 200 company. Many co-directors believed that 'cell' phones would never be inexpensive enough to permit everyone to own and use one. Wrong!!!!

Your and Cringely's words sound a tad similar.

Too, evolution of this 'cell' chip we think will be exponentially more rapid than previous technologies. We believe IBM, Sony, and Toshiba will use this technology themselves to do some radically (previously) 'impossible' stuff.

And lots of folk believe $ volumes on this widget will blow all prior records away. (Probably just more hyperbole...)

On power issues, about 30 years ago we wrote assembler code for RCA 1806 uP military systems app's. One was a PDA, 20 years ahead of its time. 1806 was a ~1MHz cpu at less than 1 Watt using then contemporary technology.

What this tells us is that IBM's et al's. 'cell' chip has an architecture which will permit, wagging, 1000 say 0.01 Watt cpus running at much slower clock rates but whose aggregate is comparatively large. That's just an old scaling notion. But these younger folk have lots of other cards up their sleeves. Transmeta is one of them. Another is unclocked evolutionary processing. Also scaling will be all of: intra, inter, and extra up and down in and out and among.

For us, this is commencement of a genuinely novel, non OO, non formal, almost quantum revolution.

But we could be wrong,


02-13-2005, 07:00 PM
But we could be wrong
It's quite likely that at least one of you is wrong.

02-16-2005, 08:51 PM
From Nicholas Blachford (stated by Craig R. Arko above):

"The first Cell based desktop computer will be the fastest desktop computer in the industry by a very large margin. Even high end multi-core x86s will not get close. Companies who produce microprocessors or DSPs are going to have a very hard time fighting the power a Cell will deliver. We have never seen a leap in performance like this before and I don't expect we'll ever see one again, It'll send shock-waves through the entire industry and we'll see big changes as a result."

That's a pretty bold statement to make. Who knows if he is right or not... but nevertheless, it's a pretty bold statement to make. I guess the proof is in the pudding as they say...