I recently read The Race for a New Game Machine, a new book which details the trials and tribulations behind the creation of the chips (which run on the POWER architecture, hence the title of this post) which powered Microsoft’s Xbox360 and Sony’s Playstation 3 next-gen gaming consoles.
The interesting thing that the book reveals is that the same IBM team responsible for designing the Playstation 3 chip (the Cell) with support from partners Sony and Toshiba was asked halfway through the Cell design process to adapt the heart of the Playstation 3 chip for the chip which would go into Microsoft’s XBox360 (the Xenon)!
Ironically, even though work on the Xbox360 started way after work on the Playstation 3’s chip, due to manufacturing issues, Microsoft was able to actually have a test chip BEFORE Sony did.
As the book was written from the perspective of David Shippy and Mickie Phipps, two the engineering leads from IBM, the reader gets a first-hand account of what it was like to be on the engineering team. While the technical details are more watered down than I would have personally liked (to be able to market this to the broader public), the book dove a lot deeper into the business/organizational side of things than I thought IBM legal would allow.
Four big lessons stood out to me after reading this:
- Organization is important. Although ex-IBM CEO Lou Gerstner engineered one of the most storied corporate turnarounds of all time, helping to transform IBM from a failing mainframe company into a successful and well-integrated “solutions” company, Shippy and Phipps’ account reveal a deeply dysfunctional organization. Corporate groups pursued more projects than the engineering teams could support, and rival product/engineering groups refused to work together in the name of marking territory. In my mind, the Cell chip failed in its vision of being used as the new architecture for all “smart electronic devices” in no small part because of this organizational dysfunction.
- Know the competition. One thing which stood out to me as a good best practice for competitive engineering projects was the effort described in an early chapter about IBM’s attempt to predict how Intel’s chips would perform during the timeframe of the product launch. I’m not sure if this is done often in engineering efforts, but the fact that IBM tried to understand (and not undersell) the capabilities of Intel’s chips during the launch window helped give the IBM team a clear goal and set of milestones for determining success. That their chip continues to have the highest operating clock speed and a throughput computing power which vastly exceeds Intel’s high-end chips is a testament to the success of that effort.
- Morale is important. If there was one feeling that the authors were able to convey in the book, it was frustration. Frustration at the organizational dysfunction which plagued IBM. Frustration at not quite ethical shenanigans that IBM played in to deliver the same processing core to two competitors. Frustration at morale-shattering layoffs and hiring freezes. It’s no secret today that IBM’s chip-making division is not the most profitable division in IBM (although this is partly because IBM relies on the division not to make profits, but to give its server products a technology advantage, which then lets them sell more profitable software and services). IBM is certainly not doing itself any favors, then, by working its engineers to the point of exhaustion. Seeing how both authors left IBM during or shortly after this project, I can only hope that IBM has changed things, or else the world may be short yet another talented chipmaker.
- Move like a butterfly, sting like a bee. Why did Microsoft “get the jump” on Sony, despite the latter starting far far in advance? I trace it to two things. First, immediately upon seeing an excellent new chip technology (ironically, the core processor for the Playstation 3), they seized on the opportunity. They refused to take a different chip from what they wanted, they put their money where their mouth was, and they did it as fast as they could. Second, Microsoft set up a backup manufacturing line in Singapore (at a contract chip manufacturer called Chartered). This was expensive and risky, but Microsoft realized it would be good insurance against risk at IBM’s line and a good way to quickly ramp up production. This combination of betting big, but betting smart(with a way to cover one self if one is wrong) is a hallmark of Microsoft’s business strategy. And, in this case, they made the right call. The Xbox 360, while not performing as well as Nintendo’s Wii (which incidentally runs an IBM POWER chip as well), has still been fairly successful for Microsoft (having the highest attach rate – games sold per machine – of any console), and they had the backup plan necessary to deal with the risk that IBM’s manufacturing process would run into problems (which it did).
If you’re interested in the tears and sweat that went into designing IBM’s “PB” processing core (it’s revealed in the book that PB stands for PlayBox – an in-joke by Shippy’s team about how the technology being designed was for both the PLAYstation 3 and the xBOX), some first-hand account of how difficult it is to design next-generation semiconductor products, or how IBM got away with designing the same product for two competitors, I’d highly recommend this book.
Book: The Race for a New Game Machine (Amazonlink)