(Rumour) XB2 CPU @ 65nm

[Dave Baumann]

http://www.teamxbox.com/news.php?id=5327

Today we can confirm that the Xbox 2 CPU will also be built using a 65-nanometer manufacturing process.


“It’ll be built on a 65-nanometer process,†a source confirmed to TeamXbox. “IBM has already taped out experimental samples at its East Fishkill fab but it will take between 12 to 18 months for them to deliver commercial parts. Anyway, they’re way ahead of Intel.â€

There are countless stories as to why Microsoft decided to drop Intel in favor of IBM. But sometimes, it just could be as straightforward as Jodie Foster claims in the movie Contact, "The simplest hypothesis is most likely to be true."

And the truth is, when it comes to microprocessors, IBM has been pulling off one success after the other. It was the first company to deliver a 64-bit processor for the desktop, the PowerPC 970 found in Apple’s Power Mac G5 and, contrary to what most analysts predicted, it is the first company to deliver a 90nm microprocessor: the second generation PowerPC 970 found inside the new Xserve G5 that will also power the second revision of the G5. With this updated PowerPC 970, IBM delivers a 90nm processor before Intel’s Prescott.

“With the new 90nm manufacturing process, IBM broke the 2 Ghz barrier. The 65-nanometer technology will allow them to break the 3 Ghz barrier for sure and get closer to the 5 Ghz mark,†our source further clarified and was quick to add, “However, this is not just about clock speed. The more important thing here is what this baby and its specialized cores can do in a single clock cycle.â€

For those really technically savvy, you may have noticed the same inaccuracy that we perceived. He used the word “cores†instead of “unitsâ€, which left us wondering…is Sony the only one coming with a CPU that is made of small groups of cores working together to process tasks in parallel? Only time will tell.

Not the first time I've heard multi-core though.

 

[Tahir2]

DaveB you bad man, you post a rumour from teamxbox - I hope you are wearing flameproof clothing

 

[gurgi]

Weren't they right about a hardware rumor before?.. xbox2 going with IBM iirc.

 

[London Geezer]

*And let the flood come in!*

Bets open, how long will it take for this thread to get to 10 pages?


Anyway, back to topic. It was expectable that MS would go for multi-core architectures, if they want to keep up with PS3's hype. (see i didn't use "PS3's power".. i'm getting better at this)

 

[Dave Baumann]

Well, I did label the post as "Rumour". However, regardless of where these things are from you have to look at the content and evalutate if there is actually any sense.

The longer the release time goes out the more likely 65nm is to be used, and I would assume that IBM would have some knowledge on 65nm by then! And, as I said, multi-core is something that I've heard linked to in regards to the XB2 a number of times.

 

[Brimstone]

Well if the CPU ends up being produced on a .65nm node, then I would assume the ATI's VPU will get the same treatment. This would result in the VPU being a generation beyond the PC version of the R500. This is assuming that IBM's Fishkill plant is going to be producing chips for Microsofts next X-Box console.

I'm awaiting the day for the confirmation if the CPU is based off of a traditional Power PC CPU or the much hyped "CELL".



I also wonder how much bandwidth Microsoft is going to get for the CPU and VPU. This slide from Micron shows GDDR-4 being available by 2006.

Hard OCP Article

 

[London Geezer]

Well if the CPU ends up being produced on a .65nm node, then I would assume the ATI's VPU will get the same treatment. This would result in the VPU being a generation beyond the PC version of the R500. This is assuming that IBM's Fishkill plant is going to be producing chips for Microsofts next X-Box console.

I'm awaiting the day for the confirmation if the CPU is based off of a traditional Power PC CPU or the much hyped "CELL".

(It's not going to happen, but) think if both Xbox2 and PS3 are Cell-based.... Now that's gonna confuse a lot of people... Impossible, or at least it would not be that easy. We already know that Ati will be in Xbox2 therefore the graphics parts of the 2 consoles will be diametrically opposed (If PS3's VPU is Cell based like we have heard in the past, that is)

 

[Fafalada]

Well, I did label the post as "Rumour". However, regardless of where these things are from you have to look at the content and evalutate if there is actually any sense.

Well there is sense, but ultimately my opinion is based off the suggested launch window I heard about. Not gonna go there right now though

My real reason for posting in this thread is this though

contrary to what most analysts predicted, it is the first company to deliver a 90nm microprocessor: the second generation PowerPC 970 found inside the new Xserve G5 that will also power the second revision of the G5.

So if it's only 5months behind it still qualifies as first?

 

[V3]

Multi-cores seems to be the big hint from some people on this board.

If that's the case like Fafalada said, it should be fun watching Deatmeat do 180

 

[Deadmeat]

..

I am not sure if MS is going to fab its chips at IBM. IBM is a high-cost fab compared to Taiwanese and Japanese foundries and this is why IBM had trouble competing with them in the past.

MS/IBM agreement is a design license agreement and not a manufacturing agreement, MS is free to take the IBM supplied design and fab it elsewhere.

But at least it appears that XCPU2 is indeed a Power5-- and not PPC980, by the implication of multicore...

I'm awaiting the day for the confirmation if the CPU is based off of a traditional Power PC CPU or the much hyped "CELL".

CELL is incompatible with NT kernel while Power5 wouldn't have a trouble running it. That fact alone excludes the possibility of CELL.

 

[marconelly!]

I think it might be something in between. A multi-core Power PC. Perhaps not as many cores and not as paralel processing oriented as Cell, preserving some aspects of the traditional prcessors?

 

[notAFanB]

CELL is incompatible with NT kernel while Power5 wouldn't have a trouble running it. That fact alone excludes the possibility of CELL.

it would be nice if you could at least substantiate our claim a little more than that.

I cannot see would kind of massive, insurmoiuntable obstacles prevent this. please elaborate and try and avoid the jargon.

 

[Brimstone]

Re: ..

I'm awaiting the day for the confirmation if the CPU is based off of a traditional Power PC CPU or the much hyped "CELL".
CELL is incompatible with NT kernel while Power5 wouldn't have a trouble running it. That fact alone excludes the possibility of CELL.

How can you exclude any possibility when a company as powerful as Microsoft is involved? Time is on Microsofts side this time around and a huge market is at stake.

 

[Dave Baumann]

Re: ..

I am not sure if MS is going to fab its chips at IBM. IBM is a high-cost fab compared to Taiwanese and Japanese foundries and this is why IBM had trouble competing with them in the past.

There is probably a good chance that MS will need someone to do the layout as well as the fabbing, which I've not really associated the Taiwanese foundries with - NEC has done for console vendors in the past, but I'd suspect that IBM will also be quite happy to take money in order to do this.

 

[London Geezer]

I said it from the beginning and i'll say it again, the real winner in the next generation will be IBM (who will be in PS3, Xbox2 and N5<---[rumour but u never know]). And ATi too (considering they're going to be in both N5 and Xbox2)...

 

[marconelly!]

Well, it probably won't be "CELL" as such, as Cell is co-op project of Sony, Toshiba and IBM. I'm pretty sure Sony has a big say there, and they probably wouldn't allow third party to start using it at the same time as they do, even that name is most likely trademarked. Now, it might be some kind of Cell derivative though, something using simillar philosophy...

 

[Deadmeat]

...

I cannot see would kind of massive, insurmoiuntable obstacles prevent this. please elaborate and try and avoid the jargon.

Simple. Memory limitation.

Power5 is a shared memory SMP machine; it runs one copy of operating system and addresses are visible by all processors.

CELL is a split memory "share nothing" MPP machine, but implemented on single piece of sillicon. Every APU gets its own exclusive memory space which is fairly small, not big enough to support something as large as NT. Both the kernel and process has to fit within 128 KB size for optimal performance, which is obviously impossible with NT but possible with some special purpose kernels.

How can you exclude any possibility when a company as powerful as Microsoft is involved? Time is on Microsofts side this time around and a huge market is at stake.

Because Microsoft is a software company and its strength comes from the hundreds of millions of lines of code it has accumulated, not hardware. Can't run NT? Forget it.

 

[Deadmeat]

...

I'd suspect that IBM will also be quite happy to take money in order to do this.

MS will shop around to bargain for best pricing. The economy is bad and lots of hungry fabs would want a piece of Xbox biz.

 

[Panajev2001a]

Re: ...

I cannot see would kind of massive, insurmoiuntable obstacles prevent this. please elaborate and try and avoid the jargon.

Simple. Memory limitation.

Power5 is a shared memory SMP machine; it runs one copy of operating system and addresses are visible by all processors.

CELL is a split memory "share nothing" MPP machine, but implemented on single piece of sillicon. Every APU gets its own exclusive memory space which is fairly small, not big enough to support something as large as NT. Both the kernel and process has to fit within 128 KB size for optimal performance, which is obviously impossible with NT but possible with some special purpose kernels.

Crap Deadmeat, I see how much you have read of the patent....

The Kernel does not have to reside in the APUs' Local Storage, but in the "Shared" DRAM... come on, even you agreed that the OS was run by the PUs which process data from the Shared DRAM and their L1 cache.

 

[Deadmeat]

...

The Kernel does not have to reside ikn the APUs' Local Storage,

Which NT requires. I was explaining why CELL couldn't run NT.

but in the "Shared" dRAM...

Which I undestand as pipes to interconnect various microprocesses spawned by the parent process.

CELL is not a multithreaded uni-process machine; rather it is a single threaded multi-process machine.