Re: Execution Timing


"Dr Ivan D. Reid" <Ivan.Reid@xxxxxxxxxxxx> wrote in message
news:slrnfam2f4.18s.Ivan.Reid@xxxxxxxxxxxxxxxxxxxx
Wade asked a question about timing executions; unfortunately a slip
of the fingers marked his post as read and I can't recall it until I
either catch up the group or it appears on google groups...

The execution times I reported in the referenced thread actually
come from the seti@home site -- they are keeping track of the numbers, not
me. In general under Windows I keep an eye on things using Task Manager
for checking the larger picture (that's how I know my securicam is taking
up 4% of my processing power on this machine).

When I do program up code snippets for checking runtimes, I'll either
do it under cygwin on this box, or ssh into the Linux box, because I don't
have Windows compilers to hand. On either I'll use the "time" command to
look at the overall times (that's what I did in the other referenced
thread)
or put specific calls to e.g. CPU_TIME() to profile specific areas of the
code.

One could also use gprof to further profile code timings, but I've
not had cause to do that (except once when I started to work on the seti
code, but real life soon intervened...).
Not much of the thread to which he refers very fortran-like, but
optimization appears to be similar one syntax to the next. Les provided
this very informative link:
http://www.flounder.com/optimization.htm

They had another link showing a computer landing a plane. I can't find it.
I just wanted to respond to Dr. Reid and allow comment on fortran aspects of
the thread that is now half the traffic in this forum.

I have always suspected that Ivan Reid is the Nom du Net of John Reid of
MR&C fame. Ivan and John are the same name, and they both live in greater
London. Anyways.
--
Wade Ward


.