| captain carot said: @Curl: There is no integer SIMD. That is why i set it in brackets. As well as limited FP SIMD. @Alby da Wolf: IBM shifted its focus with Power architecture. And at that time Apple was the only customer left that needed refreshs on an at least yearly basis. At the same time Apple did not sell enough to justify the cost of those new designs. They where as well just one of many customers IBM had. So yes, from IBM's perspective they where a minor customer with very high demands. Intel already built their chips for anyone who bought them, in terms of PC-market that meant something like 3 to 4% Mac with IBM Power and the rest being x86 with a major part of it being Intel. IBM's major customers all where to be found in high performance computing and buying complete systems. The rest was like stying with one design for several years. I as well dont believe Macs would be as successful as they are today if they still had IBM Power instead of x86. Apples switch made it easier for many devs to make Apple versions of their software. As for Jaguar, that is quite enough for such 'slow' GPU's to have balanced systems. Wii U is a little different. More CPU power, meaning more cores and clock, would have made ports easier. Let's just once take AC IV as example. That game is very likely CPU-limited on Wii U. And i am pretty sure it could have been optimized better. But the version as is might have run with rock solid 30fps if Wii U had a 1.6GHz quadcore instead of the 1.24GHz triple core. |
Yes, I agree with your points, but I'd add a few things:
As we can deduce from your points too, the main problem between IBM and Apple was plans and policies far too different and distant. The problem never was PowerPC, or even less POWER, being underpowered intrinsically, but just PPC last mainstream implementations, due to IBM focusing on higher segments, too expensive for Apple needs, and obviously preferring a far less frantic product range refresh. With hindsight, IBM current open HW licensing policies could have brilliantly solved the problems for both IBM and Apple, as Apple could have been a far more appealing customer for some POWER licensees, while IBM would have continued receiving the image return for POWER in the consumer market from Apple with very little hassles, just assisting in product design (obviously not for free) without having to actually produce the chips, with that fast production process refresh implied that IBM so hates. This way, depending on a chip maker that would have needed it far more than IBM, Apple could have received the CPUs it desired, so this way Macs could have been as successful as with the switch to x86 (let's not forget that in the golden times of PowerPC, Macs were so desired also by buyers not ready to spend Apple prices, that, right or wrong choice as it could be, Apple started licensing Mac to clone makers).
