Hi David

Thanks for the test.
I also tried with a Croquet-3.8 vm and discovered that if I wait a few 
seconds between runs the time is always nearly the same, but if I run 
repeatidly the script without waiting (typing several do-it in workspace 
window) the runs are really much faster ... what makes me think about 
commonSend/methodCache that could be disturbed by another squeak process 
(interesting).
I would like to know a direct and simple method to profile it (adding  cache 
hit / cache miss counters to the vm and rebuilding the vm?).
Any comments - ideas - whatever are welcome
:o)

"David Corking" <lists at dcorking.com> a écrit dans le message de news: 
AANLkTikZY3TuCkgkXmpMGPHK3WevZMAR55a4QUf4U-kr at mail.gmail.com...
Alain_Rastoul wrote:

> Did you (or someone else) reproduce this or is it my laptop ?

No - I could not reproduce it.  I tried with a recent Trunk image and
a Squeak 3.8-derived image (Etoys) and consistently got times within
10% to 20% of each other.   I used an Intel dual core, with fairly
recent 32-bit Mac OS X builds of the virtual machine:

Value of
SmalltalkImage current vmVersion
in these two instances is

 'Squeak3.8.1 of ''28 Aug 2006'' [latest update: #6747] Squeak VM 4.2.4b1'
'Squeak3.8.1 of ''28 Aug 2006'' [latest update: #6747] 4.0'

Hope that helps.