Book: The Race for a New Game Machine

AzBat

Agent of the Bat
Legend
The Wall Street Journal posted an article about a new book titled "The Race for a New Game Machine: Creating the Chips Inside the Xbox 360 and Playstation 3". It's authored by David Shippy and Mickie Phipps, who are IBM engineers that worked on both CPU projects. The article from the WSJ decided to focus on the negatives for Sony, which was even a little too much for this Xbox fan. The article is linked below. So be forewarned.

http://online.wsj.com/article/SB123069467545545011.html

As for the book, it's a hardcover published by Citadel, is 256 pages long & is list priced at $21.95. You can purchase a copy from Amazon for $14.93...

http://www.amazon.com/Race-New-Game-Machine-Playstation/dp/0806531010

I'd like to see reviews from people here who have read the book.

Tommy McClain
 
The Wall Street Journal posted an article about a new book titled "The Race for a New Game Machine: Creating the Chips Inside the Xbox 360 and Playstation 3". It's authored by David Shippy and Mickie Phipps, who are IBM engineers that worked on both CPU projects. The article from the WSJ decided to focus on the negatives for Sony, which was even a little too much for this Xbox fan. The article is linked below. So be forewarned.

http://online.wsj.com/article/SB123069467545545011.html

As for the book, it's a hardcover published by Citadel, is 256 pages long & is list priced at $21.95. You can purchase a copy from Amazon for $14.93...

http://www.amazon.com/Race-New-Game-Machine-Playstation/dp/0806531010

I'd like to see reviews from people here who have read the book.

Tommy McClain
This could be a very interesting read. I might get it.
 
I'm certain that a number of posters from B3D would be interested in this book and I may give it a shot after I dwindle down by current book list. My only concern is whether the writers write entertainingly. That said, I'm sure it will give great perspective and possibly shed light on other matters.
 
Does anybody have any insight as to whether the statements in the article are true?

It outlines a course of events I wasn't aware of.
 
From the article, it looks like an entertaining read but it looks like it does so by spicing up and adding a tinge of controversy where there really is none.

"Mr. Shippy and Ms. Phipps detail the resulting absurdity: IBM employees hiding their work from Sony and Toshiba engineers in the cubicles next to them; the Xbox chip being tested a few floors above the Cell design teams. Mr. Shippy says that he felt "contaminated" as he sat down with the Microsoft engineers, helping them to sketch out their architectural requirements with lessons learned from his earlier work on Playstation."

There is nothing absurd about this situation. IBM, Sony and Toshiba were working on shared technology at the time, but MS was an IBM (only) client. What would have been absurd is IBM engineers giving Toshiba and Sony engineers access to MS tech for no other reason then curiousity or MS, Sony, IBM, Toshiba all happily designing and testing their chips out in the open in the same room.
 
It also seems to suggest that the main reason for the PS3's 1-year 'delay' on IBM being unable to produce the chips fast enough.
 
Are they saying all R&D money Sony spent making the CELL leaked out into MS?

aren't the cores in the waternoose the same as the core in the ps3 ? I was under the impresion that the cores in the waternoose just had some enhancments
 
The WSJ is one of those publications you can consistently count on to be 'Sony negative' in tone when it reports, but well it is what it is.

Anyway, that said I knew that Sony had to be inadvertently paying for XeCPU development, and that at least some tech related to the PPE had to be traveling upstream, because the similarities of those cores was always way to convenient from where I was sitting.

I don't think the Cell (alone) can be blamed for the delay though, since we all know BD optical drives were essentially non-existent as well. We sort of know already everything the article provided, it's just that we now have (sort of) a human element associated with one aspect of those times.

Based on the content of the article though and the length of the book, I question the depth we'll get into the subject; I have a feeling it will be very focused on primary sources within MS and IBM and likely deal with more the net effects of the plays run. I might get it myself though of course.
 
Are they saying all R&D money Sony spent making the CELL leaked out into MS?

Not exactly all. But Sony's R&D spending seems to have contributed to the creation of the 360's CPU and some of the knowledge acquired has been shared with MS as well for their project since they were clients that wanted a similar product.

It is natural that IBM would share knowledge and material acquired to negotiate what their client wanted and what was best for both interests
 
I enjoy these kind of books.

I was very satisfied with "The Xbox 360 Uncloaked" as well as "Opening The Xbox" before it, so I'll have to check out this one.
 
I was very satisfied with "The Xbox 360 Uncloaked" as well as "Opening The Xbox" before it, so I'll have to check out this one.

Xbox Uncloaked I think is going to be a little more substantive than this one, which I get the feeling might play a little more towards imaginations and lust of intrigue.

Are they saying all R&D money Sony spent making the CELL leaked out into MS?

Well as far as R&D dollars, only the stuff that went into the PPE core would have been involved here, and that IP would still technically fall under IBM's dominion as an extension of POWER. The SPEs and the EIB which are really at the heart of what Cell is about are clearly off-limits, since those had Toshiba and Sony directly wrapped up in them in terms of IP ownership. So I imagine IBM was sort-of boldly shopping out the PPE and tweaks to it (remember that IBM wanted to go all PPE essentially with the STI project to begin with), and whatever analysis of its performance vs the Cell was all wink-wink, nudge-nudge under the table type stuff, because it would have had to have been.

On the one hand MS obviously did coast to an extent in terms of the sweat built up in development... and we've read from other reports/interviews that they went the automation route for die/transistor optimizations and such, reducing time from design to fabrication. But honestly, even if that's a sort-of coup on its own level to have benefited from that research, MS might have been better off if they had pursued a completely different processor design altogether. The XeCPU is successful (if it is that) due to its simplicity when compared to Cell's complexity, not because it's actually a processor that would win any awards. So I'm sort of dubious of the WSJ's portrayal of the situation as a coup on MS' part; what it is though is a flashlight on IBM's internals when it comes to related/competing projects like these. You'd think that with the money Sony and Toshiba were pouring into a whole new architecture of which IBM was also a beneficiary (indeed today the primary beneficiary even), not to mention the money Sony invested into East Fishkill, that IBM might have felt some bonds or something. And I'm sure on some level, it was a complete clusterf*ck of NDA's, corporate firewalls, and mixed agendas.
 
Last edited by a moderator:
. And I'm sure on some level, it was a complete clusterf*ck of NDA's, corporate firewalls, and mixed agendas.

From reading the article, that is what sounds like the entire book is about. Which is incredibly interesting on its own right in terms of corporate dealings and such.

Not being a hardware engineer, this is the first time that I heard that the PowerPC chip that was used in the 360 and Apples was based in anyway off technology that was the basis for Cell and paid for by Sony's R&D.

I was under the impression they were completely different architectures and for some reason I had though that the PowerPC architecture was around in Apples before the 360 was even in development.

So, I'm a bit confused. Is the Cell just an 'extension' of the PowerPC chip? Adding the SPE's on top of the basic design?

I'm also under the impression that the entire Cell processor's design (purpose/function) has been completely scrapped and essentially, the entire cost of developing the chip is a waste for all those involved when the development of other chips (core i7 for example), have vastly out performed anything the Cell processor was supposed to do. Especially considering the entire 'distributed computing' model has gone out the window which was supposed to be the key component of the chip.

Am I just completely wrong? I know I have just a very basic understanding of the engineering and architecture of the CPUs.
 
Not being a hardware engineer, this is the first time that I heard that the PowerPC chip that was used in the 360 and Apples was based in anyway off technology that was the basis for Cell and paid for by Sony's R&D.

I was under the impression they were completely different architectures and for some reason I had though that the PowerPC architecture was around in Apples before the 360 was even in development.

Well certainly you must have heard before the similarities between the PS3 PPE and the XeCPU cores; it's been oft discussed around here. Now to square away with the obvious firewalling that should have been in place, the previous assumptions were that each respective team simply drew from the same original R&D project at IBM as the basis of their PPE designs. I forget the name of it, but it was of a high-speed prototype from 2000 or somesuch.

Apple chips, the 970, G4s/G5s... none of those are related to what we see in these consoles, these designs are based on a different branch.

Now so the question is whether both teams did draw from the same original R&D project, or whether the PS3 team drew from it, and then the 360 team derived from what the PS3 guys were doing.

So, I'm a bit confused. Is the Cell just an 'extension' of the PowerPC chip? Adding the SPE's on top of the basic design?

The Cell chip is deterministic behavior, individual SPE processors, high-throuput, and the EIB slaved to a Power PC core. That's sort of what you said yourself, but I'm changing the emphasis to reflect what the chip is about. They could've done Cell without the PPE core at all for instance; could just as easily have had a different controlling architecture there, or none at all if design had shifted. I think the authors of this book though were closely linked to said PPE project, given the emphasis.

I'm also under the impression that the entire Cell processor's design (purpose/function) has been completely scrapped and essentially, the entire cost of developing the chip is a waste for all those involved when the development of other chips (core i7 for example), have vastly out performed anything the Cell processor was supposed to do. Especially considering the entire 'distributed computing' model has gone out the window which was supposed to be the key component of the chip.

Not at all! Has the Cell become a commercial success? Certainly not, and the whole consumer electronics/Cell scene I hear the wind blowing period (though Toshiba will finally show something in 2009 it seems). But from a design perspective I would consider it a qualified success (though I guess it depends on your design goals). The chip is even today still a leader in several of the areas in which it competes - basically against GPGPU in the cost/performance HPC space. In those areas - signal processing, cryptology, physics, etc - I would still rate the Cell more highly than the i7, especially from an architectural perspective as you get near linear scaling with additional nodes.

It's hard to know what place the Cell architecture will ultimately be given in the history books, and whether the architecture will evolve after this generation or be retired/supplanted, but it has earned some important "firsts" in the industry either way. Above all, I think the design philosophies of its creators have been fully vindicated given the current arc of computing. Even if it dies after this gen, I think it'll always have been a case of certain 'near misses' - like should they have just gone cache instead? - but at least for some folk, I think it will always endure as a pioneer that had many novel things to bring to the table.

For IBM especially I believe, even if they change their arc significantly, I think they truly lucked out having Sony/Toshiba convince them to go this route, as it gave them a credible commodity chip with which to address the areas I listed above. They find themselves in a very advantageous position now infrastructure/SDK-wise relative to where they might otherwise have been in this new era that would have seen GPGPU ascendant and Larrabee on the horizon regardless of whether they were a player in that field or not. It's just too bad for Sony, Toshiba, and the Cell itself that that PPE was the PPC core that went into the chip, as truly it is holding back its appeal.
 
It seems that Cell was designed to be more multifunctional to be used in a variety of devices where Xenon was designed specifically for console only.

I wonder if the developers that felt "contaminated", were cutting/adding things that they knew specifically were relevant for games from the limited knowledge of what the Cell road map was going to be.

I know it's been discussed in the revitalized X360 VMX threads, but maybe the changes to VMX128 were done knowing what Cell SPEs were projected to become.

It seems like MS made a really smart decision in going with IBM. I remember when the X360 first launched about how fast Xenon came to be (or what it Xenos??). Apple also must have seen that Cell wasn't going to be a successful as they chose Intel.

I find it disappointing that Cell didn't have a bigger impact on the industry. With Sony, Toshiba and IBM behind the technology, I'd have thought it would had made much bigger in roads. I seem to remember that it would be especially relevant to medical imaging technology.

Does the complexity for programming for Cell have anything to do with its lack of market penetration? Or is the complexity limited to game development?
 
It seems that Cell was designed to be more multifunctional to be used in a variety of devices where Xenon was designed specifically for console only.

Cell had different fields in mind, and XeCPU I would say had no field in mind; Flops was the obsession and MS tried to amp them up. If you read early on (pg 1) in the VMX thread you reference, you'll see that there are reports that MS wanted an entirely different kind of chip but IBM couldn't deliver - namely one with OOE. That would have been fundamentally better for MS in all likelihood, VMX and Flops notwithstanding.

I know it's been discussed in the revitalized X360 VMX threads, but maybe the changes to VMX128 were done knowing what Cell SPEs were projected to become.

Of course.

It seems like MS made a really smart decision in going with IBM.

I would say the opposite here, but oh well. :)

Apple also must have seen that Cell wasn't going to be a successful as they chose Intel.

Well since the PPE sucked and the power envelope was ass, yeah... Apple walked away from it. If IBM were at all smart or focused, then what they would have done is put effort into developing an Apple-worthy PPE and subbed that into the Cell rather than the other way around. Might have led to a drastically different outcome.

Does the complexity for programming for Cell have anything to do with its lack of market penetration? Or is the complexity limited to game development?

If it had more penetration, more people would be familiar certainly, but it just simply operates along a different paradigm entirely than a lot of programmers are used to operating within in terms of coding practices.
 
It was 'smart' with MS going with IBM in the sense that they probably knew they'd get residual benefits from Sony working with IBM on Cell and also Xenon is a much simpler to code for.
 
It was 'smart' with MS going with IBM in the sense that they probably knew they'd get residual benefits from Sony working with IBM on Cell and also Xenon is a much simpler to code for.

Maybe from a strategy standpoint it seemed smart, but the Cell PPE and the XeCPU cores are so horrid that if that's MS' reward, I'm sure they're welcome to it. The XeCPU isn't easy to code for because MS "won" these terrible cores, it's easy to code for because it's a traditional homogeneous multi-core layout with MS' dev tools on top. If MS had just gone Intel again that chip would have been just plain better in my opinion. But of course 'ownership' and IP is a lot of what led MS to IBM to begin with from Intel. But at least they have parity of sorts in that core processing that takes place, and I guess that's worth it's own sort of pat on the back.

Ironically, MS' 'acquisition' of the PPC tech might actually have been more a benefit to Sony that MS in the long run, as it allows for easy/basic porting from 360 to PS3. If the 360 had been rocking on a modern OOE x86 processor, ports to PS3 from 360 could have been quite cumbersome/burdened... likely to Sony's detriment, as general parity in 3rd party support is to their definite advantage right now.
 
Maybe from a strategy standpoint it seemed smart, but the Cell PPE and the XeCPU cores are so horrid that if that's MS' reward, I'm sure they're welcome to it. The XeCPU isn't easy to code for because MS "won" these terrible cores, it's easy to code for because it's a traditional homogeneous multi-core layout with MS' dev tools on top. If MS had just gone Intel again that chip would have been just plain better in my opinion. But of course 'ownership' and IP is a lot of what led MS to IBM to begin with from Intel.

Yeah understood however just imagine if xenon was just as hard to program for as cell for 3rd parties. I'd rather MS stayed with Intel as well, would have made BC probably easier. Hopefully both companies will make smarter deciscions in the next gen, even if they are going the 'evolutionary' route with their CPUS. I'd hope they fix what was the biggest short comings.
 
Back
Top