For an octacore to offer tangible benefits over a quadcore in terms of minimum framerates, this would effectively mean that you would have to have extremely well balanced load on cpu threads, which is something that is extremely difficult to achieve.
I believe this is difficult when we're talking about something like an FPS game. There is a lot of difficulties in threading the work because you're going to have to sync it all back to give you a consistent experience. However for the most part these problems are tackled at engine level, so although they are hard, once the industry cracks the problem it'll be a done deal for the most part. (This may have currently been done, but it takes a while for the game dev's to catch up.)
Yes there may be a few exceptions to the rule, such as games where the primary cpu drain is AI (e.g. SupCom) as that lends itself very well to distributed threads, but games with heavy AI processing are usually SP games rather than competitive multiplayer as we are talking here.
Actually the funny thing is the whole genre of games like that would appear, at least from my armchair spectators point of view, to be a whole lot easier to deal with than an FPS game would be. The funny thing is, most of the current gen RTS games haven't updated their engine in a long while, and thus are some of the worst offenders for relying on ST performance.
When you factor in that consoles currently don't have a huge number of cores this will further dissuade massive investment in engine optimisations for more than 4 cpu cores over the next couple of years. I just can't see us having a situation where BD will consistently be giving better min fps in competitive online gaming until at least 2014 by which point chances are something better will have come along anyway.
From my understanding, platform specific parts of the engine should be abstracted, thus it's entirely possible that you could buy the next ID engine and have it take up a different number of threads dependant on the platform. This however was not my point.
Today, and possibly in the future, you'll probably have one thread that does a disproportionate amount of work when looking at all threads. The real question is, if giving a dedicated core to work on, is that thread fast enough? Once the answer to that is yes, you need not worry about ST performance, providing the CPU doesn't need to share the work in a way which drops the performance of the main thread below an acceptable level.
My point is not that I believe bulldozer will give me better minimum FPS today, nor do I happen to think this will be the case in the short term future, I just happen to think it's possible it'd give me enough minimum FPS whilst giving me additional benefits in other areas, thus making it's personal value greater to me.
So the real question is what is more likely in the near future, requiring more ST performance for your main thread, or requiring more threads of deal with stuff like advanced AI and PhysX? If I go back to your previous argument, consoles games may stop us adding additional threads, but they're unlikely to require us to get better ST performance either now are they?
