Xenon and Revolution CPU's

[mkillio]

Has there been any word on Xenon's and Revolutions CPUs? I haven't been following next-gen info to much, but I'm finally starting to get interested. Any chance of throwing an altered AMD FX64 in one of them?

[Joshua Luna]

I believe both Microsoft and Nintendo have deals with IBM for their CPUs.

[pc999]

Acording to DeanoC XeCPU is a clean sheet design.

[Megadrive1988]

both CPUs are from IBM using POWER and/or PowerPC technology.


the Xenon CPU is almost certainly a multi-core CPU. The supposed leaked Xenon documents indicate 3 cores on one chip. but at least one other report said "four or more" PowerPC cores.

http://www.gamesindustry.biz/content_page.php?section_name=dev&aid=2897


The Xenon CPU is also said to be a "clean sheet" design.


the Revolution CPU is also said to be multi-core. although according to some reports, there are two different versions of the Revolution hardware in development. one of these has a single CPU / single core CPU, if I recall. less is known about the Revolution CPU than the Xenon CPU ( that is, *if* reports & docs about Xenon are true) but since the Revolution is meant to offer backwards compatibility with Gamecube, somehow Revolution must be able to run Gekko code, at least I would imagine. I doubt Revolution will have a seperate Gamecube chipset to offer BC, so I'd imagine ATI & IBM hardware for Revolution will be able to run Gamecube software through the new chipset.

now of course, we will have to see if the reports, documents, and other speculation about Xenon and Revolution CPUs match up to what the actual final hardwares have.

[Guden Oden]

The Xenon CPU is also said to be a "clean sheet" design.

HIGHLY doubtful.

For starters, starting from scratch would be pointless, as all you'd be able to do would be to come up with something similar to what's already been developed. Second, it would cost entirely too much money. Third, it would take WAY too f'n long (takes YEARS to finish a modern CPU design); if MS and IBM started the second they were finished with the original xbox they still would not be finished with the CPU. As we know, IBM has not been involved that long, so hence, they could not have started with a completely clean sheet either.

the Revolution CPU is also said to be multi-core. although according to some reports, there are two different versions of the Revolution hardware in development.

We don't know anything about revolution's CPU. We don't even know for sure it's being done by IBM, as neither have issued a press release to that effect. The "some reports" saying there are two versions of revolution in development is pretty much equal to "some guy on the internet talking outta his butt" and nothing else. :P Nintendo's too smart to waste money on developing two pieces of hardware when only one's going to be used.

but since the Revolution is meant to offer backwards compatibility with Gamecube

Is this officially confirmed? I don't think so. Besides, it sounds unlikely to me as Nintendo has never done anything similar before. Besides, there would be little to no benefit in going to such lengths anyway as people don't buy new consoles to play old games. They buy new consoles to play NEW games.

[Qroach]

HIGHLY doubtful.

I suggest you listen to the man. He knows far more than you'd probably like to acknowledge. IBM offer services where they will custom design a CPU based on your requirements.

[pc999]

The Xenon CPU is also said to be a "clean sheet" design.

HIGHLY doubtful.

For starters, starting from scratch would be pointless, as all you'd be able to do would be to come up with something similar to what's already been developed. Second, it would cost entirely too much money. Third, it would take WAY too f'n long (takes YEARS to finish a modern CPU design); if MS and IBM started the second they were finished with the original xbox they still would not be finished with the CPU. As we know, IBM has not been involved that long, so hence, they could not have started with a completely clean sheet either.

the Revolution CPU is also said to be multi-core. although according to some reports, there are two different versions of the Revolution hardware in development.

We don't know anything about revolution's CPU. We don't even know for sure it's being done by IBM, as neither have issued a press release to that effect. The "some reports" saying there are two versions of revolution in development is pretty much equal to "some guy on the internet talking outta his butt" and nothing else. :P Nintendo's too smart to waste money on developing two pieces of hardware when only one's going to be used.

but since the Revolution is meant to offer backwards compatibility with Gamecube

Is this officially confirmed? I don't think so. Besides, it sounds unlikely to me as Nintendo has never done anything similar before. Besides, there would be little to no benefit in going to such lengths anyway as people don't buy new consoles to play old games. They buy new consoles to play NEW games.

As I said acording to Deano it really is a clean S.D.

And Nintedo had said that it will be BC.

[Farid]

I suggest you listen to the man. He knows far more than you'd probably like to acknowledge. IBM offer services where they will custom design a CPU based on your requirements.
I think that what Guden calls doubtful is the possibility that IBM create a three core CPU, literally, from scratch.

IBM making a Multi-core CPU by using, and tweaking one of their already existing cores is more likely, for me too.
Also, I don't think Deano meant that the XeCPU had nothing in common with past IBM architecture, by the way.

[Qroach]

It's not just that, each cpu in that multicore is a custom design created by IBM and MS engineers. MS picks what cpu features they want, and IBM implements them.

[DeanoC]

I suggest you listen to the man. He knows far more than you'd probably like to acknowledge. IBM offer services where they will custom design a CPU based on your requirements.
I think that what Guden calls doubtful is the possibility that IBM create a three core CPU, literally, from scratch.

IBM making a Multi-core CPU by using, and tweaking one of their already existing cores is more likely, for me too.
Also, I don't think Deano meant that the XeCPU had nothing in common with past IBM architecture, by the way.

Arghhhhh.

I've said all I can on the issue, if you don't believe me then your have to wait for more information to be publically released.

It has some things in common with past IBM architectures, same as Cell does. I.e. they share bits of an old ISA.

[Shifty Geezer]

And Nintedo had said that it will be BC.
When? All I've heard is Revolution ISN'T a replacement for GC, Ninty are trying to win round a new gaming audience with something that doesn't need the hand-eye skills of current games, and they're scrapping Pad+Buttons. Without pad and buttons, how can you play existing GC games? Unless it's something wierd like mind control?

[one]

I suggest you listen to the man. He knows far more than you'd probably like to acknowledge. IBM offer services where they will custom design a CPU based on your requirements.
I think that what Guden calls doubtful is the possibility that IBM create a three core CPU, literally, from scratch.

IBM making a Multi-core CPU by using, and tweaking one of their already existing cores is more likely, for me too.
Also, I don't think Deano meant that the XeCPU had nothing in common with past IBM architecture, by the way.

Arghhhhh.

I've said all I can on the issue, if you don't believe me then your have to wait for more information to be publically released.

It has some things in common with past IBM architectures, same as Cell does. I.e. they share bits of an old ISA.

Didn't you call XeCPU a "clean sheet design"?
"has some things in common with past IBM architectures" is too broad and diverged from "clean sheet design", and ISA is hardly related to "clean sheet design" either. Your definition of "clean sheet design" is maybe different from others? :)

http://www.ibm.com/news/us/2003/11/031.html
Microsoft has announced that it has entered into a semiconductor technology agreement with IBM. Under the agreement, Microsoft has licensed leading-edge semiconductor processor technology from IBM for use in future Xbox® products and services to be announced at a later date.

According to Bernie Meyerson, IBM Fellow and chief technologist for IBM's Technology Group, the new Xbox technologies will be based on the latest in IBM's family of state-of-the-art processors.

The agreement is a success for IBM’s chip business and an affirmation of IBM’s leadership as an innovator in advanced consumer electronics.

Microsoft and Xbox are either registered trademarks or trademarks of Microsoft Corp. in the United States and/or other countries.

The names of actual companies and products mentioned herein may be the trademarks of their respective owners.


http://eetimes.com/consumer/showArticle.jhtml?articleID=51200450
The chance to start a processor design with a clean sheet of paper doesn't come often. But IBM fellow Jim Kahle is getting his shot at revolutionizing the way computing systems work.

As chief architect of the Cell processor, Kahle is among the handful of designers at top chip makers worldwide who are working to bring parallelism into devices designed for the consumer sector.

Back in 2000, engineers from IBM Corp., Sony Corp. and Toshiba Corp. sat down to discuss a possible collaboration. IBM had been among the first to get to a gigahertz-class processor, and the engineers around the planning table were "looking at more traditional organizations of machines," Kahle said. "But what we found was that they didn't give us the computational efficiencies that our partners-Sony and Toshiba-needed."

Then the trio went to that proverbial clean sheet, drawing upon the symmetric-multiprocessing experience within IBM. The planning team "looked at the whole gamut of how to get to new levels of efficiencies," said Kahle.

Or if you're in IBM, the chance to start a processor design with a clean sheet of paper does come often, every time IBM accepts an offer from a game console maker?

[pc999]

Arghhhhh.

I've said all I can on the issue, if you don't believe me then your have to wait for more information to be publically released.

It has some things in common with past IBM architectures, same as Cell does. I.e. they share bits of an old ISA.

Probably a silly ? but what is ISA :?:

[randycat99]

Maybe they mean "clean sheet design" in the same way that the PS2 EE is a "clean sheet design"? :P :D

[Fox5]

I believe both Microsoft and Nintendo have deals with IBM for their CPUs.

IBM does fab AMD cpus.

However, I doubt an athlon cpu will make it into a console. Probably too expensive, plus AMD really hasn't made a stake for the console market since the first xbox, and they're probably unable to produce enough chips.

Plus, the advantages of the integrated memory controller would probably be minimized in a console that doesn't have all the legacy stuff of a pc, can have the memory controller on the gpu(and consoles are more gpu focused anyhow), and have the memory controller customized for specific purposes. Plus, amd would probably have to redesign the athlon 64's memory controller for a console which would take a lot more time and effort than amd has available.

Probably a silly ? but what is ISA

I'm guessing Industry Standard Architecture or Instruction Standard/Set Architecture.

[Inane_Dork]

When you said "clean sheet" I thought you meant it was totally original. I'm shocked to learn that both PowerPC and X2 CPU have ADD instructions. How bogus does that make your claim? "Clean sheet" indeed. :roll:




;)

[jvd]

it will have a higher sustained flop rating than the cell chip in the ps3 will have

[Brimstone]

Or if you're in IBM, the chance to start a processor design with a clean sheet of paper does come often, every time IBM accepts an offer from a game console maker?

Getting nervous about the Xe CPU and its potential realworld performance numbers? Let me wipe that sweat from your brow, Merry X-mas.

[DeanoC]

Probably a silly ? but what is ISA :?:

Instruction Set Architecture.
A family of processor share certain instructions (the encoding in the program stream), in theory sharing an ISA lets processors execute the same code if you only use those instructions in the shared ISA (there are usually extensions which aren't compatible).

x86 is an ISA, a Pentium 4 isn't in any related to 8086 but they share an ISA.

x86-64 is an ISA, both AMD-64 and Intel 64 bit processors use it.

POWER is an ISA, as is PowerPC (they are closely related but not compatible).

PowerPC is used in lots of systems, the base instruction set (there are actually a number of versions) is compatible across them. How they execute those instructions can be radically different but the basic ISA is the same. Its seems likely the all three processors in next-gen consoles will use this ISA (its public for Cell in the ISSCC abstracts, the others aren't official but considered very likely).

Of course this doesn't mean that they will be compatible in any way.

[pc999]

Thanks

BTW they can do a totaly new CPU to this right ISA ,i.e. , a clean sheet design, Right?

[Panajev2001a]

it will have a higher sustained flop rating than the cell chip in the ps3 will have

???

Both get their performance from VMX-style units, from parallel processing using custom SIMD Vector Units...

What is your reasoning behind that statement ?

The efficiency of SPU in PSOne ?

I seriously think you are again downplaying Sony hadware way too much.

[DeanoC]

[Probably a silly ? but what is ISA

I'm guessing Industry Standard Architecture or Instruction Standard/Set Architecture.

Thanks, BTW they can do a totaly new CPU to this right ISA ,i.e. , a clean sheet design, Right?

Absolutely.

But sometimes an ISA can be a problem to do an efficient design as certain instruction can only be done using slower methods.

This happened on POWER, they removed certain instruction that made supercalar processing much harder. The cleaned modern version would be called PowerPC.

The other way of mantaining compatibility is simple not to bother supporting these instruction directly. In most older ISAs, modern processors do some instructions via microcoding, these are 'emulated' in the processor. Are generally much slower but mantain compatibilty. Things like the old string instructions in x86 (they have even older providence going back to the 8080 processor).

[pc999]

Thank you very much.

[MrSingh]

it will have a higher sustained flop rating than the cell chip in the ps3 will have

???

Both get their performance from VMX-style units, from parallel processing using custom SIMD Vector Units...

What is your reasoning behind that statement ?

The efficiency of SPU in PSOne ?

I seriously think you are again downplaying Sony hadware way too much.

sorry pana, you know it's going to be NV30 all over again!!!

merry xmas! :D

[Megadrive1988]

I have little doubt that the PS3 CPU will have a higher peak flops performance as well as a higher sustained flop performance than Xenon's CPU. no reason not to think so, at this point. unless Cell and PS3-CPU is some sort of massive disaster, which is pretty unlikely, even if it doesn't reach the lofty levels that have been 'floating' around :) no pun intended

[Wunderchu]

I have little doubt that the PS3 CPU will have a higher peak flops performance as well as a higher sustained flop performance than Xenon's CPU. no reason not to think so, at this point. unless Cell and PS3-CPU is some sort of massive disaster, which is pretty unlikely, even if it doesn't reach the lofty levels that have been 'floating' around :) no pun intendedI agree

[one]

Or if you're in IBM, the chance to start a processor design with a clean sheet of paper does come often, every time IBM accepts an offer from a game console maker?

Getting nervous about the Xe CPU and its potential realworld performance numbers? Let me wipe that sweat from your brow, Merry X-mas.
"potential" and "realworld" sound mutually-exclusive to me :D

XeCPU's ISA will be compatible with PPC970 or a subset of PowerPC instructions just like RISC-oriented POWER4 (from which PPC 970 was derived) ISA, plus SIMD units backward compatible with AltiVec. Why I assume it compatible with AltiVec is because without it you can't test optimization on Power Mac G5 devkits. If the compiler can take care of everything and Xenon developers don't have to care too much about its platform, devkits could be x86 Windows systems but actually Mac were used. (Well Microsoft might want it as a beta-test of Windows PPC kernel but who knows) Except for AltiVec just trimming compatibility-related things from PPC970 can give it performance gain. In addition to ISA, you can cut little-endian support and 32-bit support. But I don't call such a treatment a "clean sheet design".

Gekko for the GameCube lacks AltiVec but has Write Gather Pipe, a data compression function and a floating point pipeline, but it's still not called a "clean sheet design" AFAIK.