Q: What makes Morphic slow?
Stefan Matthias Aust
sma at 3plus4.de
Sat Nov 20 18:43:14 UTC 1999
At 10:56 19.11.99 -0500, Bob Arning wrote:
>One technique often overlooked is using MessageTally.
Thanx for that hint. It's a nice tool - but is there any documentation for
the MessageTally available? I'm not sure how to interpret the results.
Being one of the old-pentium owners IMHO morphic is still slow on current
hardware.
The expression "Browser openBrowser" needs about 1.1 seconds on my old P100
and feels even slower - probably because the screen update is defered.
A spy log reveals
**Leaves**
5.6 StringMorph(Morph)>>layoutChanged
5.6 Point>>+
[...]
3.7 UndefinedObject>>ifNil:
3.7 Rectangle(Object)>>ifNil:
[...]
which I read as that you could probably save about 7% of execution time by
inlining the ifNil: method. It's also interesting that #layoutChanged is
called that often. Unfortunately, each time I run the test, I get
completely different result - even if I start always with a blank screen.
So can you really trust in the result? How long should a sample run to get
meanful numbers?
bye
--
Stefan Matthias Aust // Bevor wir fallen, fallen wir lieber auf.
More information about the Squeak-dev
mailing list
|