More on Macs, MHz myths, and madness

Looks like it's time for a follow-up to my Mac post from the other night. The rant was linked from Slashdot today, and our server went off the deep end, while my inbox filled with flames. Plus, there are some other developments. Among them:
  • First and foremost, I am looking for a loaner Macintosh G4 866MHz system for use in a TR-style benchmark face-off (like so). I would like to settle this outright, like we do the AMD-Intel disputes, by testing like hardware boxo-a-boxo. A number of readers have requested it, and I've always been willing to do the test. If you think you might be able to help, drop me a line.

  • Next, the guy behind the Mac-vs-PC chart that got me riled up in the first place has updated his chart, bowing under the weight of criticism. You've gotta respect that. In his updated chart, the Mac G4 system is beaten on points by a Pentium 4-based system from Gateway. Of course, the chart's still daffy, but it's nice to see the update.

    He has also revised his statement claiming a 1GHz Athlon was equivalent to a 1.7GHz P4 and an 866MHz G4, removing the Athlon. (We've shown an Athlon 1.2GHz is closer to the P4 1.7GHz.) Further, he's admitted a mistake in omitting an Athlon-based system from the chart, and promises to include one next time around. If this keeps up, the PCs will be winning by fifty points in no time! :)

  • For those of you weary of coming here and getting an unresponsive web serever, take heart. Our new, dually Linux box is getting close. Once we've got it up and running, we should be able to weather traffic storms with much more grace.

  • Finally, I have a bit more argumentativeness to get outta my system. I can't count how many Mac guys have come through here and dropped a comment or an e-mail message saying "MHz don't matter; it's the MHz myth" and leaving it at that—as if clock speeds were as useless as colored cases, and as if there weren't objective standards of performance by which such things could be compared. This is where the Apple marketing machine just boils my blood. I really have nothing against Macs or G4s, but they are, quite simply, slower than current high-end PCs. To claim otherwise is to deceive, and to do so with reference to the "MHz myth" is to feed a calculated, corporate-planted misconception about how things work.

    Here's the comment I posted on the subject:

    OK, Mac freaks. I will say this once clearly. Listen up. Read and comprehend.

    MHz matters. It's one part of overall processor performance. There is a marked difference in performance between a 400MHz G4 and a 500MHz G4. Or between a 700MHz Athlon and a 1400MHz Athlon. This is a fact.

    The other big variable in the equation is IPC, or instructions per clock. IPC * clock speed = overall performance. (Roughly. Other components in the system, like the front-side bus and cache/memory architecture will boost or hinder IPC.)

    All of these things are well known and understood in the PC enthusiast world. The Pentium 4's IPC is lower than the Athlon's. Likewise, the G4 "Plus" executes fewer instructions per clock than the older, 500MHz-and-lower G4.

    Not all instructions are created equal, of course, so IPC * MHz isn't the whole story. That's why we have benchmarks.

    Overall, the G4, even in its higher-IPC form at 500MHz, isn't much faster clock-for-clock than its x86 competition. Check out zamboni's SETI numbers, in which the CPUs process fast Fourier transforms, to get a feel for the match-up:

    Of course, these things will vary according to the type of work being done, processor optimizations, and the like. To see a fairly nuanced look at how a pair of processors stacks up when clock speeds, compilers, and the types of math vary, I suggest you read this article.

    This is why I say the Macolytes have ignored advances in the PC realm for the past year or more. You guys just don't get it. We aren't buying any "MHz myth." We are simply buying the reality that MHz matters, as one part of overall performance. Sadly for Apple & co., the G4's clock-for-clock performance isn't strong enough to overcome the disparity in clock speeds. Not even close.

    Spare us the lectures about myths until you have actually grasped the realities.

And that's about enough for now.
Tip: You can use the A/Z keys to walk threads.
View options

This discussion is now closed.