mprof v1.0 - a simple Matlab profiler.

To use these files for profiling Matlab code, do like this:

Initially, copy mprof to a directory in your $PATH and copy ctic.m and
ctoc.m to a directory in your matlabpath.

Each time you want to profile an m-file, e.g. calc.m, do like this:

Execute 'mprof calc.m c.m' at your unix prompt. Now c.m contains all
code from calc.m interleaved with calls to the profiling code.

Do the following inside matlab:
>> c(your_ordinary_calc_arguments)
>> global CPU_TIMES

CPU_TIMES(i) will contain the time spent at line i in calc.m.

Suggestions:
>> find(CPU_TIMES==max(CPU_TIMES))
will give you the number of the line taking the longest time to
execute.
>> [y,i]=sort(-CPU_TIMES)
will give you the line numbers in i sorted in descending order by the
total time
>> bar(CPU_TIMES/sum(CPU_TIMES)*100)
will give you a percentage graph

KNOWN BUGS:

Doesn't handle multi-line assignment, e.g.
A= [ 3 4 5;
3 6 7];

For feedback, send an email to niclas@cs.umu.se.

Have fun!

/Niclas
--
Niclas Borlin                       email: niclas@cs.umu.se
Department of Computing Science     phone: (+46) 90-166832     
Umea University                       fax: (+46) 90-166126    
S-901 87 Umea, SWEDEN            lat/long: 63.49 N, 20.19 E