More cellfun related benchmarks

[Rik]

8/6/11

All,

The previous test results I posted showed that cellfun is horribly slow
when the function used is defined in an m-file.  Further testing shows that
there is also a difference between passing a function name as an input or a
function handle.

Benchmarks:

cellfun with string "sin"
1.0773

cellfun with function handle @sin
1.2197    +13% slower

cellfun with anonymous function @(x) sin (x)
1.4817    +37% slower

cellfun with "isempty"
0.010728

cellfun with @isempty
1.2066    ~100X slower

If the function happens to be one of the specialized ones that is directly
implemented by cellfun then the speedup is over 100X.

All of this suggests that cellfun should always be called with a string
argument since it will be faster than the equivalent function handle and
can potentially be much faster in common instances.

Accordingly, I went ahead and made the change for the core Octave functions
in this changeset
(http://hg.savannah.gnu.org/hgweb/octave/rev/cefd568ea073).  Running 'make
check' as a crude benchmark showed there was at least a small improvement
from the change.

Cheers,
Rik