04 January 2006

Myth or Reality #3: PERFORMANCE

The thinking here is that the new (Intel-based) machines will be significantly faster than comparable (PowerPC-based) machines. Paul Murphy's blog (which originally got me thinking on these myths/realities) doesn't directly make this point - in fact he implies that "delays" to Merom and "Woodrow" will be a problem for Apple and suggests that in fact newer PowerPC variants (dual-core G4, and low power G5 chips) could be used in the higher end machines. But, I wonder a little about whether Paul is missing something here. I've heard of Merom (laptop market), and I've heard of WoodCREST (server market). I've also heard of Conroe (desktop market). These have been slated for 2nd half 2006 release, and I am not aware of reported delays at this time.

Anyway Merom/Conroe aside, back to the point of this post, performance is another area I've been watching with interest. There have been many anecdotal reports of Mac OS X on Intel being hugely fast, but I really wonder about this. For instance, the dual-core, dual-cpu PowerMacs seem to fair up pretty well next to comparable Wintel boxes today in real world tests. While many claim the PowerPC G4 is way behind, I wonder how much it really is (I've seen XBench scores on G4 machines that are not much different form similar GHz G5's). If Mac OS X on Intel massively outperforms PowerPC-based machines, AND Apple were to release the new machines in the order predicted (ie low-end machines first) then it is almost certainly going to obsolete even it's high end (and high-margin) machines very quickly. If the Mac mini is 2x faster than today's mini, then it would change the argument considerably for buying an iMac. More so in the laptop arena. Why would anyone buy a PowerPC-based Powerbook when a cheaper Intel-based iBook beat it hands down? And, Apple would be in danger of seriously undermining it's PowerMac computers if they showed up really badly against a machine a fraction the size and cost. No matter how many minis Apple sells over the next few quarters (even if they do have the same margin) it's not going to make up for a complete dive in Powermac/Powerbook or even iMac sales. This all leads me to conclude that the first Intel Macs will perform solidly - perhaps very well on a few tasks, but not outperform the higher end PowerPC models. Of course, Apple could restrict performance by keeping many of it's i-applications as PowerPC apps and letting Rosetta drag the performance down a bit. But I don't think Apple would play that game for that reason.

Nevertheless, Apple's conundrum is that it needs to show that Mac OS X on Intel outperforms Windows on Intel (it will be a problem if it lags well behind), and demonstrating that a rationale for the move (which was not the explanation given at the time) is indeed performance. But not at the complete expense of decimating sales of it's high end machines or indeed making it look foolish for not moving much sooner.

My take on this is that the first Intel macs will NOT be super-fast machines running rings around more expensive PowerPC based machines at any task thrown to them. This actually implies lower-end Yonah chips (use of the single core variant, with a concentration on the lower voltage models perhaps). There will be a few nuggets of super-positive performance numbers to whet the appetite of the faithful that once dual core 64bit Merom is about, then these babies will fly. And there'll be some positive numbers to show that Mac OS X stacks up well against Windows itself. Apple will flirt by showing a little ankle but not opening the Kimono quite yet. (Apologies for the metaphor, must have been thinking about Memories of a Geisha or something!)

Score a draw here - part myth on day 1 but reality by later in 2006 and early 2007.

No comments: