Re: 1,342,532,544 combinations of algorithms

[Bill Sommerfeld <sommerfeld@east.sun.com>]

> And yes, I realize that the function call overhead may be trivial
> compared to the overhead of DES, so arguably this coding style is
> taking things too far...

When I've gone down this particular rathole, I've found that
optimizations of this form make a bigger difference than you might
expect at first, particularly if you want to squeeze the last 5% or
10% of performance out of a system.  Some of it may come down to
"memory is slow" issues.  In one case, at a previous job, eliminating
a data copy and tuning up the byte swaps in an md5 implementation on a
big-endian risc system made a noticeable improvement in performance.
[Upon arrival at my current job, I discovered my new coworkers had
discovered the same thing independantly.]

					- Bill