Will partnering with Intel give Apple a computer faster than a PC?
I can foresee a future where the best engineers at Intel move to the Apple project. Being able to start with a blank slate for x86 will probably be the most interesting job over there.
For the past 25 years or so the x86 has been forced to be backwards compatible back to the original MS-DOS. This includes all aspects of the architecture, from the instruction set, to the memory system, and even the BIOS. If you made an x86 machine without this backwards compatibility, could you do better? I think this is the question that Jobs is posing to Intel right now. Ironically, this new system would probably still run Linux, but running stock Windows might be difficult. It’s hard to say how much of that backwards compatibility is required for Windows to boot, let alone run programs.
At the current x86’s core is a RISC chip with a big huge layer of crud on top if it that emulates all the CISC instructions. That layer could be trimmed. The current chipsets contain tons of legacy support that could be jettisoned, as Intel has shown with tons a prototypes that were never picked up by the industry because of Windows users who demand that Lotus 1–2–3 still runs on it. The form factor of the board is completely open for discussion as there need not be a standard case that accepts it. So if they really take advantage of the situation I could see a Mac running on Intel that is faster than the fastest Windows PC.
That being said I do worry about the relative rates of hardware delivery between Apple and Dell. Apple has very long hardware cycles while Intel ships new chips, especially at the high end, quite often. It could easily turn out that Apple machines are always slower than the fastest Dell box. This could be even worse if Apple decides to rely on a custom Intel chip and chipset. Basically the gains from such a move would have to be huge for it to be worth it. In this situation they would basically have to ship a machine that is much faster than a stock PC, then just as the PCs are catching up, rev their hardware. I find it unlikely however cool it would be.
BTW, does anyone really think this will make high-end Macs cheaper? It looks like they will be more expensive to me. For instance, the processor I would like to see in a machine that came out tomorrow would be the Pentium D 840 EE (HT, dual core, 64-bit) is around $1200. I’m not sure how that compares to 2 2.7 ghz G5s, but I would think it is more expensive.
Update: Paul Thurrott has replied to this entry on his blog, since he doesn’t have a comment system, I guess I’ll just talk here.
I totally agree with his sentiment regarding the Mac platform not being about performance but rather than entire user experience. However, I can’t help but note the hypocrisy of his entry when his previous entry was about Mac game performance. People will discuss it and I think it is interesting to speculate about it if only to give us something to do between now and when the first Intel Macs are released.
On the other hand, increased performance gives you the ability to offer a better experience. Take the Mac desktop as an example. Many of the interesting effects are best done in the context of extremely fast graphics cards. Without them the experience degrades, mostly gracefully, but it isn’t the same as when you have a really fast system.
People are always going to compare performance, especially people that push their systems to the limit. It isn’t relevant for the general population, only for those that really need to keep their CPU pegged most of the time. If thats not you, go ahead and buy an iMac or an iBook and don’t worry about the higher end machines.
In the meantime, I for one welcome our new Intel overlords and hope they give Apple some key advantages they can take to the desktop designers so the Mac can become ever more useful, easy to use, and, of course, beautiful.