Thank you for the very worthwhile post. Have you been considering writing a book?

It really seems the book isn't worth buying.

First, the book seems to be worthwhile for someone who's looking at trying to manage complexity within their engine better and hasn't yet had a taste of working on larger software projects or game development projects. But if you've already got a bunch of books on this subject, I think it's just more of the same.

As for writing a book on this, hahaha. Sorry, I'm afraid I don't have the patience and I'd feel too dirty doing so. I don't know enough on the subject to offer a comprehensive delivery on the materials. And besides, it's really too early for such a book. It's going to depend on what we get in next generation consoles, as that will largely drive the direction the industry will take.

Will the console manufacturers feel content in shipping new consoles with power-concious, low-cost, middle of the road performance? For example, something with 4 general purpose cores and GPU on the same die. Yeah, that'd be great to have now in consoles, but I think in a couple of years it'll be pretty mediocre for something built on a 22-28nm process node. But the console builders may very well do this to maximize their own profits, and consumers will still buy it up as it will be able deliver games slightly better than today's but at 1080p resolutions and 60hz frame rates. If that happens, it might put a stop to the need to investigate this further.

I think the magic number is 8. If we start getting 8 or more cores to work with, that will force developers to start approaching things differently.

If we're lucky, we might get that in next generation consoles, or we might get something even better. Maybe something like Larrabee. Who knows, I'm not privy to that kind of information.