Could PS3/Cell be another Itanium or Talisman?

[MfA]

Doesnt really matter if it is revolutionary or not, what matters is that it will be the only one of its kind for the near future.

 

[Spidermate]

Doesnt really matter if it is revolutionary or not, what matters is that it will be the only one of its kind for the near future.

Exactly. And for that same reason, that is what will separate it apart from everything else.

 

[randycat99]

I would say it is certainly a radical direction for a console design, if not revolutionary. If it ends up redefining what goes into a workstation or desktop PC, I would say that would be a radical direction for PC's, as well. It's pretty hard to be "revolutionary" in anything because just about everything has been done to varied degrees of success. What may not have been possible using yesterday's tech may become ravenously successful using today's tech. So the idea may not be new, but if it achieves distinctly higher levels than its conceptual predecessors, that is a good thing to acknowledge. Given that, it is usually the essence of being "radical", rather than "revolutionary", that makes something notable these days. It just so happens that the term "revolutionary" is more often used to catch the eyes of people, when technically, very few things today are truly revolutionary by virtue of all we have done already.

 

[passerby]

WARNING: CELL SPECULATION
A bit of 'Cell speculation', since this is the current 'Cell thread', so to speak.

Thinking about the comments 'it is highly configurable for different tasks'.
Thinking about the assumption of 'configuring very many cheap, simple CPUs together in a clever manner'.
Then thought about the end of the original Cell paper, which describes APUs configured as a graphics pipeline, not as a massively parallel system(so to speak).
Then Jov posted this on another thread.

Now, general purpose computing is a definite waste and expensive in consumer electronics. Plus how costly is the final product going to be?

The Guess: (not entirely original, but more elaborate)

An APU is a very, very simple chip. So simple that we may be surprised at its simplicity when it is revealed. But that's fine, because it is simple by design. Its true purpose is not to work alone, but to be configured with other APUs, in a pipeline that is specialised for the device it is to be deployed in. In this manner it is possible to deploy the same APUs on different devices that perform different tasks. The difference is that each set of APUs has a different 'pipeline configuration' - one that is customized for the specialized needs of the device it is deployed in.

Hence we get the 'Cell solution' - cheap to produce(all APUs are the same, and very simple) and highly configurable for different applications.

How it applies to PS3 - I'm guessing that it will replace the role(s) of the VUs in the PS2. It will be used to construct a programmable processing pipeline for graphics - for that is the major specialised need of a console. A possibility is regarding all those papers nAo posted about tile/brick-rendering/whatever, this will all be implemented in the Cell configuration. In other words, all the APUs are already configured in specialized processing pipelines to execute the tasks described in those papers.

I am not discounting parallel execution - we may have more than one 'APU pipeline group' executing simultaneously.

My guess also seems to make an APU unsuitable for executing AI code. A possibility is that there may be 'CPU-subprcessor APUs', like the PS2's VU0! If an APU can be just sufficiently more advanced than a VPU, maybe the APUs can handle more complex work than a VPU, leaving the CPU with more resources to spend.

OK, I'll keep quiet now and wait for end of FY 04.

 

[Spidermate]

I would say it is certainly a radical direction for a console design, if not revolutionary. If it ends up redefining what goes into a workstation or desktop PC, I would say that would be a radical direction for PC's, as well. It's pretty hard to be "revolutionary" in anything because just about everything has been done to varied degrees of success. What may not have been possible using yesterday's tech may become ravenously successful using today's tech. So the idea may not be new, but if it achieves distinctly higher levels than its conceptual predecessors, that is a good thing to acknowledge. Given that, it is usually the essence of being "radical", rather than "revolutionary", that makes something notable these days. It just so happens that the term "revolutionary" is more often used to catch the eyes of people, when technically, very few things today are truly revolutionary by virtue of all we have done already.

This sounds the most logical. Infact, something very, very similiar was quoted alone those lines. Let me note, however, that this architecture is said to be more advance than previous attempts tried by others.

 

[Brimstone]

Sun's Proximity Communication is a huge breakthough and will most likely end being more revolutionary than the "CELL architechture". Also a CPU with an asynchronous design is more radical.

In Proximity Communication, chips are positioned within microns of each other, but not necessarily touching. That permits transmitters on one chip and receivers on the other to exchange data at on-chip speeds without being connected by off-chip wires, soldered connections, or other current techniques that really slow things down.

...

"If the amount of heat you can dissipate is fixed, the only way to increase performance is to use power more efficiently," Drost says.

Which is why he finds Proximity Communication so exciting. No wires means fewer transistors using a lot less power. It also means you can easily add a whole bunch of wireless communication channels to deliver a massive amount of bandwidth.

And it offers real advantages over the approach others have taken to 3-D chip technology, which is to sandwich several chips together.

"It is possible, by stacking chips, to get quite high bandwidth between them, but it doesn't help with the next step," Drost says.

"Say you're to the point where you can laminate a few chips together. That's great, but if you don't have Proximity Communication, what do you do now? How do you get this collection of chips to communicate with another collection of chips at high speed?"

With Proximity Communication, you can do both.

"We think it's valuable to laminate a couple of chips together, where it makes sense," Drost says. "You can actually increase the density of data and processing in the system. But then the outside chip will have Proximity Communication technology on its outside face and that's how it will communicate with the rest of the system."

What Drost and other members of the team are trying to do is build a supercomputer with hundreds of thousands of processors talking to each other with bandwidth that's comparable to on-chip cache.

"It's a whole different computer when you can do that," Drost says.

http://www.sun.com/presents/minds/2004-1115/

Asynchronous design is an approach to building computer processors that allows them to be built and run in a more modular way. In microprocessor designs, for example, it allows engineers to build different components of the microprocessor -- the floating point unit or an arithmetic logic unit, for example -- in modular fashion.

This kind of design lets processors perform more efficiently than traditional, clock-based, microprocessors and also allows Sun engineers to reuse processor components in future designs, something that is rarely done in today's microprocessors, Mitchell said. "When you do asynchronous logic, it tends to use half to a third as much power as a clock system," he said.

http://www.nwfusion.com/news/2003/0818darpawork.html

 

[Spidermate]

Sun's Proximity Communication is a huge breakthough and will most likely end being more revolutionary than the "CELL architechture". Also a CPU with an asynchronous design is more radical.

In Proximity Communication, chips are positioned within microns of each other, but not necessarily touching. That permits transmitters on one chip and receivers on the other to exchange data at on-chip speeds without being connected by off-chip wires, soldered connections, or other current techniques that really slow things down.

...

"If the amount of heat you can dissipate is fixed, the only way to increase performance is to use power more efficiently," Drost says.

Which is why he finds Proximity Communication so exciting. No wires means fewer transistors using a lot less power. It also means you can easily add a whole bunch of wireless communication channels to deliver a massive amount of bandwidth.

And it offers real advantages over the approach others have taken to 3-D chip technology, which is to sandwich several chips together.

"It is possible, by stacking chips, to get quite high bandwidth between them, but it doesn't help with the next step," Drost says.

"Say you're to the point where you can laminate a few chips together. That's great, but if you don't have Proximity Communication, what do you do now? How do you get this collection of chips to communicate with another collection of chips at high speed?"

With Proximity Communication, you can do both.

"We think it's valuable to laminate a couple of chips together, where it makes sense," Drost says. "You can actually increase the density of data and processing in the system. But then the outside chip will have Proximity Communication technology on its outside face and that's how it will communicate with the rest of the system."

What Drost and other members of the team are trying to do is build a supercomputer with hundreds of thousands of processors talking to each other with bandwidth that's comparable to on-chip cache.

"It's a whole different computer when you can do that," Drost says.

http://www.sun.com/presents/minds/2004-1115/

Asynchronous design is an approach to building computer processors that allows them to be built and run in a more modular way. In microprocessor designs, for example, it allows engineers to build different components of the microprocessor -- the floating point unit or an arithmetic logic unit, for example -- in modular fashion.

This kind of design lets processors perform more efficiently than traditional, clock-based, microprocessors and also allows Sun engineers to reuse processor components in future designs, something that is rarely done in today's microprocessors, Mitchell said. "When you do asynchronous logic, it tends to use half to a third as much power as a clock system," he said.

http://www.nwfusion.com/news/2003/0818darpawork.html

Well, seeing how we don't yet know how powerful Cell is, that really can't be predicted.