Benchmarking

Slow can mean slower than it could be or slower than we would like or slower than is acceptable for that job. So the faster you make your scripting language the things it can be used for because it is fast enough for more jobs.

speed does matter but only to the extent that the choice of solution is fast enough.

>speed does matter but only to the extent that the choice of solution is fast enough.

Yes. Years ago, I spent a decade as a performance geek, and that is the one thing many people had trouble understanding. Unless you know how fast something needs to be, you never know if its fast enough.

> Unless you know how fast something needs to be

I remember when I made a change on a drawing and went out to have a cigarette cause the redraw 'might' be finished by the time I got back. only a couple of years later, that redraw would be almost done in 30 seconds, and soon after almost instantaneously. After each performance increase, we would be amazed at the increase in speed and even wonder if we could work to the new performance standard. We always managed and in a few weeks were wanting even better speed and performance. So much of this performance thing is only our expectation.

One used to see big differential gains, but as technology has progressed, we now measure the performances differences for almost any app performance in seconds, or milliseconds. For programmers, gamers, and others who really stress the hardware benchmarking may be some indicator, but that gets harder and harder to do. One of the reasons why CPU OEMs have now gone to multiple cores, stressing parallel execution and multiple tasks rather than sheer speed of execution. Most users will never use a fraction of the capability. So, for the average user, benchmarking is indeed of little practical use.

>So, for the average user, benchmarking is indeed of little practical use.

That's a big yes and no, if only because the "average user" is so hard to pin these days.

A sizeable portion of those average users do what once would have been considered sophisticated video and audio work, with rendering and compression tasks that can still use cycles at a prodigious rate.

Overall, though, I should think benchmarking is more interesting for those running servers.

> users do what once would have been considered sophisticated video and audio work

I'd suspect most who do have found 'you can't do that with the old PII' ;-)... Still, that's more about an appropriate package, including video and sound, than about benchmarking.