Running Out of Time

A recent review illustrates the problem with PCs nowadays.

This article overclocks a 3GHz PIV C to 4GHz and records the results.

What’s interesting about this review is that for the SuperPi benchmark, it pulls data from this forum survey, and shows how the results compare to that of earlier CPUs.

Let’s pull a few of those numbers:

Let’s do a little math and show how much time gets saved by upgrading from one generation to the next:

Or in sheer numbers:

Going from an early PII to a TBred saves you about five minutes. That’s enough time to go to the bathroom.

Going from a TBird to a Paly saves you a minute. If you need to go to the bathroom in that time, I hope you have a bottle by your desk.

Going from an O/Cd TBredB to an O/Cd PIV gives you eight seconds. That’s not enough time for you to get the bottle or anything else out.

People don’t perceive computer speed in terms of percentages. They perceive computer speed in seconds and minutes.

The reality is the average home and business user is running out (if he or she hasn’t run out already) of tasks that takes much time for a computer to do.

Let’s pretend you know somebody with a PII generation computer who wants to upgrade. Let’s really pretend that the love of his life is SuperPi.

You can get him pretty cheaply (or will shortly even if you don’t overclock) from seven minutes to less than one minute. That saves him a little over six minutes. He’ll notice that and be impressed.

Tell him that for a few hundred more, he can save ten more seconds, and he is far less likely to be impressed, or willing to pay for the improvement.

After that, though, how are you going to impress him? You can’t possibly save him six minutes again; the task now takes less than a minute. Just getting thirty seconds off will require a future 10GHz+ monster years from now.

If time literally means money to him, yes, that will impress him (or at least his or her boss).

But what about everyone else?

Let’s face it, for most computers users, the most time-consuming computer task they have is booting (and that’s precisely how many real people measure how fast a computer “really” is).

Human beings are actually pretty lousy at sensing time. They just don’t sense saving a second here, a second there.

What they do sense, profoundly, is the hours, days, or even weeks it takes to get used to an upgrade.

There’s a new term being used for this, “overshoot.” It means having computers that are more than fast enough for a given task.

The old canard “the software will drive the hardware” no longer works. There hasn’t been any killer app requiring a killer computer for the average person for years, and there are none on the horizon. This is not to say one will never, ever show up again, and if it does, people will do what they have to do, but not until then.

All the average person does is email, a little web browsing, a little word processing. Just what can you do to email and web browsing and word processing that requires a 5GHz processor, and what else can you get the average person to add to the list?

For the average person, if it becomes a choice between a cheap more than fast enough computer system or an expensive way more than fast enough computer system, guess what they’re going to choose? Guess what they are choosing?

Email Ed