Accelerating MATLAB Performance

45Proling MATLAB

®

Performance

That is, instead of doing this:

>> profile on; for idx = 1:100, a = a-b + c^d*log(e); end; profile report

do this:

>> profile on; myFunc(a,b,c,d,e); profile report

Where myFunc.m is a le that contains the proled MATLAB code within a function:

function a = myFunc(a,b,c,d,e)

for idx = 1 : 100

a = a-b + c^d*log(e);

end

end

2.1.6.3 Ensuring Compatibility with Real-World Conditions

A second type of proling, no less useful, is to test a wide variety of scenarios, each with

repeatable data/actions, rather than repeatedly proling the same scenario. The reason

for this is to ensure the validity of the discov