Other Sites

This is the exciting time of year when one may think about whether it's time to buy a new Macintosh. For most people, just the thought of having a new Mac creates a certain amount of giddiness. But then reality sets in, and one starts to weigh the considerable cost against the reality of a Mac that may or may not need replacing.

The next step many take is to search for speed comparisons. Perhaps, just perhaps, a realization of how dog slow the current Mac is will generate some psychological momentum.

Of course, the criteria for a new Mac is very different in the corporate world than it is for the home user. Businesses tend to think about cost versus productivity gains. And costs include not only the hardware but training. For the home user, apart from outright failures of the components, it's more subjective. For many, all kinds of considerations start to pour into a bucket, and when the bucket is full, it's time to spring.

For example, I have a 500 MHz Titanium PowerBook. (I'm writing this essay on it.) Aside from two very inconspicuous dead pixels in the upper left corner, almost to the edge, this Mac works perfectly. The modest clock speed doesn't chew up the battery, so it's a perfect computer with which to sit on the couch in the evening, watch TV, and compose a story. It would benchmark very poorly by today's standards, and yet with TextWrangler, it's a very nice Mac for a writer.

For those who stress their computer more, it's hard to resist looking at benchmarks. Benchmarks are fun to look at, and they can be especially useful when one is using a particular application, like Photoshop, where time is money. When more and more tasks require one to leave the room and make tea, it is indeed time to buy some speed. Benchmarks help with the Return on Investment (ROI) calculation.

For those who are casual users, however, the decision is more difficult. It may end up relating more to a new feature, a faster Internet connection, or the ability to run certain software that was impractical before.

How does one really know when it's time? (As I said, apart from an annoying component failure.) One rule I have been successfully using for a long time is the 4x rule. For me, a computer has to benchmark about 4x faster before I'm visually delighted with the speed up of general operations. That means the CPU, graphics card and disk access all combine to give an aggregated score of 4x what I have. To make this judgment, it's nice to have benchmark data, conducted on a wide range of computers with a standardized test. For example, the Geek Patrol, which is fanatic about benchmarks, has published performance results on over 50 Macintosh models, including the very latest Core 2 Duos. Using the Geekbench software, the performance of every Mac from an iBook, G4, 800 MHz to a Mac Pro, Xeon, 3.0 GHz was listed in a convenient bar chart format.

What I liked about this set of benchmarks was that it's broad enough to put a large number of different Macs into in perspective and obtain some historical insight into the evolution of modern CPUs used by Apple. For example, one should feel very good about Apple's move from Motorola G4s to Intel in the notebook line. On the other hand, the results point out the annoying fact that the PowerPC G5 is a very strong CPU and isn't exactly blown away by the Xeon. We knew that. Apple told us that. And so, if it's time to upgrade, it's probably going to be a PowerBook to MacBook transition.

Here is some sample data from the Geek Patrol compilation.

• PowerBook, G4, 800 MHz : 41.9
• Mac mini, G4, 1.5 GHz : 70.1
• MacBook Pro, Core Duo, 2.16 GHz : 184.9
• MacBook Pro, Core 2 Duo, 2.33 GHz : 222.7
• PowerMac G5, Quad G5, 2.5 GHz : 279.0
• Mac Pro, 2 x dual core Xeon, 2.66 GHz : 330.3

Note that the benchmark span from a really old PowerBook to the latest Mac Pro isn't even an order of magnitude. It takes a lot of time and engineering refinement to make big gains in raw computer speed. Minor benchmark speed increases, except for those engaged in intense computational tasks, has to be considered an overrated and possibly an emotional element in the upgrade decision.

Additionally, benchmarks don't even begin to tell the whole story. Other factors to consider include whether one is tied to an important Classic only application, whether the most important applications are Intel native or must run under Rosetta, whether one needs to run Windows apps fast (hence Boot Camp or Parallels), and whether an upgrade to one of the communication modes is needed -- such as FireWire 800, 1000BaseT and wireless technologies.

If these aren't a consideration, then with today's technology, and noting Moore's Law, a 4x speed up works out to about a 36 to 48 month purchase interval.

When my 500 MHz Titanium PowerBook, with only 8 MB of VRAM, wouldn't run iTunes 7, I knew it was time to retire the trusty old but otherwise nearly perfect Mac. In fact, I waited too long. The speed up factor for the latest MacBook Pro is more like 8x. (The TiBook will be handed down to my mom.) If one computes backwards 4x from Apple's latest, beautiful Core 2 Duo MacBook Pro, that suggests that if one's PowerBook is older than a 1.25 GHz G4 (FW800), it's time for a new MacBook or MBP.

Happy Shopping!