Sony funded Xbox360

[SiYuan]

The Xbox 360 has turned out to be a formidable competitor to the PlayStation 3, and Sony may have unknowingly contributed to its success. So say David Shippy and Mickie Phipps, who wrote a book about how IBM engineers quietly built the 360's Xenon processor in parallel with the PS3's Cell CPU. The Wall Street Journal has the details:

When the companies entered into their partnership in 2001, Sony, Toshiba and IBM committed themselves to spending $400 million over five years to design the Cell, not counting the millions of dollars it would take to build two production facilities for making the chip itself. IBM provided the bulk of the manpower, with the design team headquartered at its Austin, Texas, offices. Sony and Toshiba sent teams of engineers to Austin to live and work with their partners in an effort to have the Cell ready for the Playstation 3's target launch, Christmas 2005.
But a funny thing happened along the way: A new "partner" entered the picture. In late 2002, Microsoft approached IBM about making the chip for Microsoft's rival game console, the (as yet unnamed) Xbox 360. In 2003, IBM's Adam Bennett showed Microsoft specs for the still-in-development Cell core. Microsoft was interested and contracted with IBM for their own chip, to be built around the core that IBM was still building with Sony.

According to the Wall Street Journal, Sony and Toshiba had agreed that IBM would sell the Cell to other clients. However, they didn't expect IBM to sell "key parts" of the chip to a competing console maker before development was through.

In The Race for a New Game Machine, Shippy and Phipps recount situations like IBM staffers "hiding their work from Sony and Toshiba engineers in the cubicles next to them" and testing the Xenon "a few floors" up from the Cell designers. They say Microsoft ended up with its first Xenon chips six weeks before Sony, too.

http://techreport.com/discussions.x/16137

 

[WeasteDevil]

The problem is, it's all bullshit. XCPU and Cell have nothing in common other than the standard Power Processor core which Cell has one and XCPU has three, and even those are not the same thing. The idea that the XCPU has parts copied from Cell is quite simply wrong. What's laughable here is the article itself.

 

[SiYuan]

So there's not much truth to this?

 

[WeasteDevil]

So there's not much truth to this?

No, and it's not exactly news either, just the media being naughty and spinning sensationalist stories for the sake of it based on a review of a book by the chief engineer of the Cell's PPU. However, look at this article from 2005, it's not anything new that they share some technology between them, as they are both based in some form or other on IBM Power technology which is owned by IBM.

http://arstechnica.com/articles/paedia/cpu/xbox360-2.ars

The Xenon's triple-core design shares some DNA with the Playstation 3's Cell processor, so it's not surprising that it also embodies many of the same assumptions about the best way to wring performance out of the sorts of extremely large transistor counts that Moore's Curves have given the latest generation of integrated circuits. Like the Cell processor that will power the Playstation 3, the Xenon carries on the "RISC"-style tradition of trading programmer/compiler effort for hardware. In a nutshell, software writers who develop for Xenon must take on more of the burden of optimizing their code by making it explicitly parallel, and in return they get more execution hardware to play with.

When Sony unveiled the Cell at this past ISSCC, the relative lack of information on the PowerPC Processing Element (PPE) that makes up the "host" processor for this network-on-a-chip was a major disappointment. Details on the PPE were scarce, but the details that were released matched up exactly with information that had already been leaked about the IBM-designed PowerPC cores that would power the next-generation Xbox. Moreover, the very fact that IBM released so few details of the Cell's PPE at the Cell's debut was evidence in itself that the architecture would play an important role in the yet-to-be-announced Xbox 360.

Nonetheless, the quote does suggest that the Cell's designers (remember that the Xenon shares the majority of its PPE architecture with the Cell) don't consider branch prediction to be a good place to spend scarce transistor resources. It therefore seems likely that the PPE does not include the same level of branch prediction hardware as the PowerPC 970, despite its longer pipeline. (I'd be happy to be proven wrong on this count, though.)

As I said, it's not something that was not already known. I'd have to see some direct quotes from the said book to see how far the media are spinning this.

 

[WeasteDevil]

So there's not much truth to this?

Ok, someone (that shows a lot of XB360 bias) on another forum has read the book, and this is his/hers take on what was said, no quotes though.

I finished the book. Took an hour in total.

Don't buy it , its a waste of $23 bucks

The most interesting things is that Microsoft wanted 4 cores but later reduced it to 3 when they found out that weren't going to be able ot make it profitable.

IBM decided to go with Microsoft because Sony added another 2 SPU units up from the original 1/6 design to 1/8 design and that would cut into the profit margins for IBM who was supposed to produce the chip at their fab.

The original dual issue that IBM moved out for multithreading was put back in by Microsoft and since they used a common core for everything but VMX units (vector units ) a Sony engineer who worked on the original VMX units fixed a bug on the Microsoft verison of the unit which was enhanced over the Sony one. The bug didn't affect the sony one, yet he still fixed it.

OOE was removed because IBM couldn't get all 3 chips (Apple, Microsoft, Sony) done in the time frame and keep OOE. That pissed off Apple and seems to have set them on the road to Intel

Thats really all the technical stuff. Now if you want to hear about this guys few stressfull years making these chips its a good read. But not so much if you want to learn about the chips.

So, from this, both the PPE in the Cell and the 3 XCPU cores are based on the same original prototype design for a PowerPC core that was the work and property of IBM. When Sony, Toshiba, and IBM got together, and started to work on the Cell, they were using this prototype design as the basis for the PPE in the Cell, and a Sony engineer fixed some thing with it, and that was used in the XCPU, being IBM property.

The point here is, that the VMX unit on the PPE in Cell is probably never used, as those type of jobs will be given to the SPEs, so Sony would have probably wanted to take the VMX unit out altogether, but couldn't due to the legacy of PowerPC, Linux, etc. The interesting part here for me is the Apple reference. The SPEs in the Cell don't have any cache, so OOOE (out of order execution) capability was probably not very important to Sony at all. However, in a desktop environment (Apple Mac) it becomes very important indeed, and no matter how much IBM wanted to sell the Cell to Apple, the lack of that (amongst other things) is probably why Apple went to Intel. Apple would probably have liked something like XCPU with OOOE logic.

The bolded part is also interesting in that it contradicts what Kutaragi said in an interview very early in 2005, although the author of this book is probably wrong as he was an engineer on the PPE part only, not the SPE part (which is what makes Cell the Cell). It seems that 8 SPEs/SPUs was always his plan, and that for some reason IBM tried to cut that down. When he found out about it (Sony were paying for this), he made them put it back to 8.

Q: Cell has 8 embedded "SPE" CPU cores. What is the basis for this number?

Crazy Ken: Because it's a power of two, that's all there is to it. It's an aesthetic. In the world of computers, the power of two is the fundamental principle - there's no other way. Actually, in the course of development, there's this one occasion when we had an all-night, intense discussion in a U.S. hotel. The IBM team proposed to make it six. But my answer was simple - "the power of two." As a result of insisting on this aesthetic, the chip size ended up being 221mm2, which actually was not desirable for manufacturing.

In terms of the one-shot exposure area, a size under 185 mm2 was preferable. I knew being oversized meant twice the labour, but I on the other hand, I thought these problems of chip size and costs would eventually be cleared as we go along. But in this challenge of changing the history of computing, I could not possibly accept any deviation from the rule of the power of two. For example, the world of communication also has gone to the rule of the power of ten. Ethernet, which started with 10M bit/s, has gone through stages of 100M and 1G, and 10G is certain to come next. You won't go with, say, 4G just because 10G is technologically difficult. It is my belief that real technological innovation is born from such persistence.

There are also a lot of people that think that the PPE and the three XCPUs are very poor processors, an example of which is why the most powerful supercomputer in the world needs to use AMD processors in addition to Cells to work; why can't the PPE in the Cell do that work for itself? Because it's a poor general purpose processor, especially when considered next to the most recent Intel chips used in PCs. That said, Cell was never designed to be a general purpose processor.

As a final note, here's a quote from a software developer that has worked on both PS3 and XB360...

BTW..SPUs can execute up two instructions per clock cycle, and on decently written code a single SPU runs circles around a XeCPU core at any time of the day.

Now, if you take that as read, then Microsoft got the wrong end of a Sony engineer's work.

So, as I said, typical sensationalist "journalism" from a poor writer with an agenda on a slow news day.