Accelerating MATLAB Performance

358 Accelerating MATLAB

®

Performance

These static optimizations are available through two functions, which can be used inter-

changeably: gcompile and arrayfun. gcompile accepts both strings and function handles,

while arrayfun requires function handles. In practice, using gcompile directly can yield

a few percent performance improvement as it avoids some caching checks that arrayfun

incurs. There are two ways for invoking gcompile on MATLAB code. The rst is by using

the M-le convention as follows:

my_fn = gcompile('filename.m'); % compile function to a CUDA kernel

[A, B,...] = my_fn(C,...) % call the compiled function

Alternatively, one can inline function code directly to gcompile statement:

my_fn = gcompile(verbatim);

%{

function