Optimizing array accesses

[Clemens Lode]

Hello!

Is it worth to use a[x] instead of b[y][x] (with a=&(b[y][0]) ?
a is const int*, b is const int.
y is a static int and does only change once (at the beginning of the
program).

That would be one less memory access and thus faster, right?

Thanks :)
Clemens

[Ivan Vecerina]

Re: Optimizing array accesses

"Clemens Lode" <ghoul@clawsoft ware.de> wrote in message
news:bmbj1r$8gl $1@news.rz.uni-karlsruhe.de...[color=blue]
> Is it worth to use a[x] instead of b[y][x] (with a=&(b[y][0]) ?
> a is const int*, b is const int.[/color]
b is const int** or const int (*)[someInt] , I assume.
[color=blue]
> y is a static int and does only change once (at the beginning of the
> program).
>
> That would be one less memory access and thus faster, right?[/color]

It's hard for us to tell if it is "worth" it.
But yes, it will usually be faster to pre-calculate the
address of the sub-array that you will work with.
So it is a good idea, especially if variable 'a' is
given a meaningful name...

Regards,
Ivan
--

http://ivan.vecerina.com

[Jon Bell]

Re: Optimizing array accesses

In article <bmbj1r$8gl$1@n ews.rz.uni-karlsruhe.de>,
Clemens Lode <ghoul@clawsoft ware.de> wrote:[color=blue]
>
>Is it worth to use a[x] instead of b[y][x] (with a=&(b[y][0]) ?[/color]

Maybe. Maybe not. It depends on the quality of the optimizer in your
particular compiler, and on your particular application.

Check the code generated by your compiler to see what it does under
various levels of optimization. It may already do this particular trick
for you. If not, test both versions of the code and see it it actually
makes a significant difference in speed.

--
Jon Bell <jtbellap8@pres by.edu> Presbyterian College
Dept. of Physics and Computer Science Clinton, South Carolina USA