| Richard 2007-09-15, 6:55 pm |
| On Sep 15, 6:50 pm, Robert <n...@e.mail> wrote:
> On Fri, 14 Sep 2007 22:51:45 -0700, Richard <rip...@Azonic.co.nz> wrote:
>
>
>
> There would probably be one multiply on the subscript because the optimizer would use that
> value on subsequent accesses. (I wrote that before reading your paragraph below sayig the
> same thing.)
>
>
> 1. Because a timing test showed indexes are slower.
> 2. Because multiplication is now as fast as loading an index.
>
>
> Subscript was faster in my test because the subscript was optimized but the index
> apparently was not. It appears the code was reloading the index on every iteration.
>
>
> Most people who use indexes believe it. Why else would they use indexes?
I don't believe that you have surveyed anything but a tiny sample of
coders so you have no idea, except for what, a few dozen, how many use
indexes, nor what they believe, nor why they use indexes.
I have actually done a test on Linux with Fujitsu 7. Within the
perform loop it does a couple of moves totaling 4 uses of subscript/
index:
subscript comp-5 9.58
index 9.52
subscript comp 13.35
Index _IS_ faster, at least on this system. Granted it is not much
faster when the MF extension of Comp-5 is used.
So there is no reason to stop using index at all. Your claims,
methodology and assertions are crap.
|