There's way more to Intel Inside than a sticker
August 17, 2007
When Cox News reporter Bob Keefe asked Apple CEO Steve Jobs why Apple didn't participate in the "Intel Inside" sticker campaign last week, the Mac blogger community really jumped on Keefe's back. Unfortunately, everybody seemed to think it was a dumb question and ostracized Keefe to no end.
Personally, I have always wondered how Apple would respond to a sticker question like this. I already knew the general answer (I think we all did), but it was very interesting to hear Steve Jobs articulate it in a way only Steve Jobs can -- especially in light of his public appreciation and respect for Intel. It was a tightrope response that explained Apple's position without disrespecting Intel. Whether you think the question was dumb or not, Jobs' response was great, and I bet it was exactly the kind of thing Keefe wanted for his article. That makes the question pretty damn relevant if you ask me. Obvious questions are not dumb when the answer gives at least some amount of insight and perspective. You can pretty much ask Steve Jobs any question and be guaranteed a unique angle. Want to hear something interesting about toilet paper? Ask Steve.
I think the reaction to the sticker question speaks to a mixed awarness in the Mac community about just what it means to have Intel processors in our Macs. I know some Mac customers are happily unaware of what processor is in their Mac, and they're probably in the best shape among all of us! Others appreciate the benefits we've seen since going to Intel, namely improved performance, reduced power consumption, and faster product development. This is especially apparent in the laptop line, where MacBooks run away from the previous PowerBook G4 models by a large margin. But unfortunately, I think a large portion of the Mac population takes "Intel Inside" for granted whether they realize it or not.
I say this coming from the perspective of someone who was intimately aware of the pluses and minuses of the PowerPC architecture for many years. I developed and compiled code to extract the best performance from the hardware for scientific and engineering applications. I ran real-world benchmarks and wrote articles and white papers on topics like AltiVec, the PowerPC 970, and compiler technologies. I put my reputation on the line and took no end of heat from the x86 community when advocating and defending the PowerPC as a performance alternative to Intel and AMD, even though that was a tough position (until the G5 showed up, you absolutely had to convert compute intensive applications to take advantage of AltiVec to even make the PowerPC worth mentioning, and that was not always a viable option). Finally, I was in the loop with Apple, compiler vendors, and hardware developers where some frank language was used, and it became clear that we were pretty much stuck under the thumb of IBM and Motorola. If you think Macs languished under the PowerPC regime, you should have seen how it looked from down in the trenches, where Apple executives privately used four letter words when talking about the PowerPC. It was grim.
Things improved after the PowerPC 970 (the G5 by Apple's moniker) came out, and for a while we actually had something to be proud of. Between the hardware and IBM's excellent XL compilers, the G5 was a solid performer. If you could take advantage of AltiVec on top of the 970's already strong integer and floating point performance, it was usually ahead of the competition by a mile. I look back on the year or so after the G5's introduction as the zenith of Apple's PowerPC run, when things were good. But we all know what happened -- PowerPC 970 development stalled, missing Jobs' stated "3.0GHz in a year" target by a mile. And it just went downhill from there. The G5 got hotter and more power hungry with each iteration, and IBM was never able to deliver a PowerPC 970 variant for laptops. The G5 shot into stardom rather quickly and dramatically, but the fade out was pretty slow and painful.
I have to admit, when Apple announced they were moving to the Intel architecture in the summer of 2005, I had mixed feelings. I had built up a lot of experience with the PowerPC, the 970 in particular, and had invested a lot in hardware and software. While Apple did a great job making Xcode projects port easily from PowerPC to Intel and gave us the ability to cross target one architecture from the other, there was no such path for software outside of Xcode and the basic GNU C compiler subset Apple uses in Xcode. To this day, I require a G5 with IBM's XL Fortran compiler and a Mac Pro with Intel's Fortran compiler on my desk in order to support both platforms -- there is not a single commercial Fortran compiler that can target both architectures from one machine. This means twice as much work when compiling and deploying Fortran-based software. In effect, it's like supporting two platforms even though they are both Mac OS X.
While the migration to Intel has not come without an impact on pro developers, I think the Intel Macs are awesome, especially the second generation of Intel-based Macs that have been showing up this year. The performance, whether scaled by dollars or watts, is fantastic, and Intel's compiler tools are top notch, easily as good as the excellent tools IBM offered for the G5. The farther along we get, the more there is to appreciate about having Intel inside our Macs, and the migration issues and ups and downs of the PowerPC era will fade away. But like many things in life, you can't really understand or appreciate the pluses without having experienced the minuses. And that's why I was disappointed by the whole blogger-brouhaha revolving around stickers. Let's not forget what "Intel Inside" really means, and let's not lose sight of the long winding road that got us to this point. It's about way more than a sticker, guys.