[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

cl:time in 7.2

It's hard to say what might be causing these metering results, just from
the information you provide.  Nothing comes to mind as an obvious cause
of these results.  (In particular, arrays haven't gotten any slower.)
Very often, the reasons for performance results are not at all obvious,
and cannot be understood without more intensive metering.

One unusual thing I noticed from the data you provided is that your
applications seems to generate a lot of new symbols.  Does it call
gensym a lot?  Does it call intern a lot?  I also see that it calls EVAL
explicitly: does it spend a lot of time in the interpreter?

Can you find some sub-piece of the top-level function, and measure it in
both systems, in hopes of narrowing down which portion of your program
is the one taking more time?

If you meter the same thing many times in a row, how widely do the
results vary from run to run (within a single version of Genera)?

I recommend that you try using the new Metering tool in 7.2 to try to
learn more about where the program is spending most of its time.  If we
knew more about that, we might be able to come up with some better ideas
about what was going on in the 7.1 version.