"The Race for a New Game Machine"[1] is a pretty good read on the development of the Cell and Xenon[2] processors. Giving up the out-of-order seems like a really bad decision and the internal fighting at IBM, if accurate, is really sad.
A side question: What in the POWER architecture makes it hard to implement? I was told the addressing modes are complicated enough that it will always be slower and harder to create than other processors. I'm wondering if this is urban myth or has some basis in reality?
A side question: What in the POWER architecture makes it hard to implement? I was told the addressing modes are complicated enough that it will always be slower and harder to create than other processors. I'm wondering if this is urban myth or has some basis in reality?
1) http://www.amazon.com/The-Race-New-Game-Machine/dp/080653101... with a lot of articles written about the book such as http://www.wsj.com/articles/SB123069467545545011
2) https://en.wikipedia.org/wiki/Xenon_(processor)