Sorry guys, but i am very impressed with MS first foray into the console industry.
Well you're certainly welcome to be... I guess some of us are less impressed because Sony already demonstrated that a relative newcomer can succeed (and their success is arguably more impressive), leaving Microsoft's foray somewhat less impressive in comparison...
MS will change how things work in console gaming.
I'm curious, what do you expect them to change?
Currently it is aggressivly expanding into other markets (into embedded vs. MIPS/SuperH (see XBOX & various x86 MCUs (~10% market share)), low to medium end workstation (that is almost a complete takeover by now (Sun & PowerPC have single digit market penetration) and pushing aggressivly into low to medium end server markets (up to 50000$) & they are complementing this with their ARM/IA64 ISAs) while being unchallenged in its home market (mainstream personal computing).
I don't know if I'd consider Xbox an aggressive expansion into the embedded space. I guess maybe an Intel/x86 platform win perhaps, but x86 has always had a small piece of the microcontroller market (where CISC code density matters more than performance), it's just that that's primarily been filled by non-Intel x86 which in turn has been small peanuts compared to Motorola's m68k/ColdFire and various other microcontroller vendors (Hitachi, Toshiba, NEC, Sony, Mitsubishi).
As for desktop MPUs (whether they be consumer, professional workstation, or mid to high-end server), Intel has always been the dominant force there (maybe not in performance but definitely in size and influence). Being IBM's choice for their PC, and subsequently all major cloned systems, pretty much meant that economy of scale (along with M$'s dominance on the software side) would permit them to dominate that segment in all but the most specialized sectors. That pretty much meant the all the RISC competitors never really had much of a chance (including AMD and Intel's own RISC designs). The market for competitors really hasn't changed much in that area with the exception of AMD actually providing competition with Intel on the high end to some degree. There's still non-Intel low-cost x86 solutions around (Via and Transmeta instead of Cyrix and AMD). High-end wise, even with the relative improvements of McKinley over Merced, it hasn't done all that well (hence it's more affectionately known nick 'Itanic'). As far as their ARM solutions go, it's probably been a total blessing the DEC's demise dropped their ARM7 based work (StrongARM) into Intel's lap. While StrongARM did fairly well as a high-end solution, I don't know if Xscale (ARM10) will do the same. While ARM moves on (ARM11) and other ARM vendors compete with high-end Xscale offerings (like Samsung), not to mention their use of Xscale in comms where it's dominated by Motorola and IBM's PowerPC offerings. The real market for ARM is in the low-end where Intel doesn't really compete (I'd even speculate that Nintendo sells more ARM devices than Intel does).
Intel by now even seems to be fairly ahead of IBM on process technology (look at their ramp @.13 micron & their plans of having .09 mainstream (for prescott) at the end of next year).
I'm not so sure that necessarily makes Intel 'ahead' of IBM on process technology. IBM's CMOS8xx (.13µm) has been around longer than Intel's and CMOS9xx (90nm) will likely be available as well next year, not to mention they're solidly on their way to 65nm with CMOS10xx. That also neglects to mention their SiGe BiCMOS processes for mixed signals ICs as well (although there are rumors of Intel making a big push with 90nm SiGe, that remains to be seen). Then there's Toshiba who's already sampling .10µm devices (Sony makes the verification equipment for it), and has pretty solid plans for 90nm, 75nm and 55nm... In fact 90nm is a goal for a lot of people next year (foundries, embedded, memory, etc)...
Just to come full circle for me the x86 camp Intel & AMD seems stronger than ever, for all major competitors failed to really introduce a serious competitor, that could profit from similar economics of scale. (I still have hopes for IBMs 64Bit Power4 mainstream derivate but i feel it will be too little too late by 2004). Toshibas & IBMs microprocessor bussineses are just drwafed by Intels & AMDs nowadays.
Well considering IBM made x86 the defacto for the PC, nobody really could challenge it unless you wanted to make x86 processors yourself and the only one who's really given Intel a run for their money has been AMD. RISC vendors never really had a chance because there was no popular software platform for them to flourish (other WindowsNT for MIPS, Alpha and PowerPC of which the Alpha version only lasted any decent amount of time, and are now all gone aside from the rumored existence of Win2k builds for Alpha maintained internally). Linux and BSD of course change that (and perhaps gives IBM's PPC970 some path to popularity although I doubt it as both IBM's AIX and Linux solution will likely be too expensive for general consumers and Apple will probably be the best chance for a regular end-user to get their hands on it), but in the end they've come about too late to have any *real* influence.
As for IBM and Toshiba's microprocessor businesses. I'd say they're fairly comparable in size considering while semiconductor business represents only portions of each companies business while Toshiba is twice the size on Intel and IBM three times... I'd attribute Intel's visibility to it's core market being the more 'glamorous' desktop MPU and core-logic categories...
Is this why the PS2 is so much more powerful then the XBox?
Hehe, well considering I first saw GS hardware in late '98 and EE hardware in early '99, when Xbox wasn't even a paper spec, I'd say it does a fair job of competing...
I don't consider Intel part of the XB2 equation, they had very little to do with the original XBox. nVidia will more then likely handle the design and hand it off to a group almost assuredly including former SGI engineers.
Another interesting approach that nobody has mentioned is an all Intel solution. While it goes against what I previously argued about Intel's business patterns, considering the amount of production capability they spending on (while mainly to kill AMD, it'll be 'too much' once AMD would be dead so perhaps they plan on providing foundry services for others as well), they could possibly build a totally custom solution. It's not out of the realm of possibilities, as Intel does have the 'know-how', large-scale reliable fab-space, core-logic design experience. Perhaps something like Banias, although with more emphasis on higher performance and embedded graphics than I/O...
I am in complete agreement with you statements about diminishing returns on the visual end(artists are likely to be the key to the best visual next gen), I am interested in hearing your take on how much computing power you think it will take to significantly improve the state of AI.
Well it's hard to determine since AI has gone from "as long as it doesn't hurt frame-rate" to gaining enough significance to be critical component of a game's engine (in some cases being *the* feature basis for a game). Basically meaning that even without processor improvement, AI is benefitting from gaining a larger share of existing processor cycles. Also there's the shift away from academic methods (genetic algorithms, neural nets), to more traditional methods that yield 'more' for less cycles. Of course like any other aspect, AI can consume vicious amounts of cycles if you're willing to go there (just look at all of the computer vs. human chess competitions).
I only mention AI (as there's other aspects to consume processing cycles) as it's one of the larger determinants of gameplay and in spite of popular belief I don't think online gameplay is the be all to end all...
Oh well, enough rambling...
