Profiling the interpreter from within

[Skip Montanaro]

    gb> The extension, I hallucinate, would turn on statistical profiling
    gb> using the same interrupt mechanism used to get the times reported by
    gb> gprof. It works by setting a timer to go off 100 times per
    gb> second. On each interrupt it examines the return address and
    gb> increments a counter in an array indicating a "hit" on that address
    gb> (actually range of addresses). By adding up the counts for the range
    gb> of addresses included in a function you can get an estimate of the
    gb> time spent in that function.

In theory, that would work okay, however, since the python interpreter is so
much slower than real hardware, I think you'd have to have much longer runs
to make the statistical sampling very accurate.

-- 
Skip Montanaro (skip at pobox.com)
http://www.mojam.com/
http://www.musi-cal.com/