Why is CELLFUN(@isempty) much slower than CELLFUN('isempty')?

The title says it all. Example code:

x=cell(10000000, 1);
x(1980) = {1};

tic; find(~cellfun(@isempty, x)); toc
Elapsed time is 9.189388 seconds.

tic; find(~cellfun('isempty', x)); toc
Elapsed time is 0.061950 seconds.

Answer 1

[per isakson]

See cellfun – undocumented performance boost

[Despinida]

Thank you!