MathGroup Archive 1992

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: Timing

  • To: mathgroup at yoda.physics.unc.edu
  • Subject: Re: Timing
  • From: Boss Man <BOSS at noaapmel.gov>
  • Date: Thu, 15 Oct 1992 16:39 PDT

Mark Stabb writes:

>After all the recent talk of timings on various machines, I ran

T>iming[10000!]

>on my Mac IIci and got about 124 sec which seemed in good agreement
>with the other reported times for this configuration. Running it a
>second time however gave a time of about 1.67 sec. Clearly something
>is being done differently the second time around. What is it, and is
>everyone who has been talking timings lately been talking about the
>same ones? I especially was surprised by the recent IBM RISC 6000
>report of on the order of 2 sec.
>Mark Stabb

Wmen Mma evaluates a function,  it keeps the value around for further
reference.  The first call to 10000! evaluates the function.  The second
call simply retrieves the value claculated previously.  You are right.  
This is an important behavior to know about for making performance tests.
For the record, all of the tests I referred to earlier need to be done 
with a freshly launched session of Mma.

Ed Boss
NOAA/PMEL






  • Prev by Date: Timing[ ]
  • Next by Date: Developing Mma frontends for NeXT
  • Previous by thread: Timing[ ]
  • Next by thread: Developing Mma frontends for NeXT