Eric Scharff [mailto:Eric.Scharff at Colorado.EDU] writes:

> The feeling that Squeak 3.0 feels sluggish (especially when compared to
> 2.8) seems to be verified by the benchmarks.

I'm not sure I trust the benchmarks for this.  I see more sluggishness than
a 10% change in benchmark numbers would suggest.

My configuration is a Celeron 450 running Windows 2000.

I was trying to pin down a repeatable, quantifiable test, but didn't get
anywhere with the quantifiable part.  Here's what I did:

Open a fresh Squeak 2.8 image (with either 2.8 or 3.0 VM).   Enter a morphic
project. Open a browser.  Now click around on lots of classes and methods.
Feels fast.

Open a fresh Squeak 2.9 or 3.0.a0 image (with either 2.8 or 3.0 VM).  Get
that browser up.  Click on lots of classes and methods.  On my machine,
there is a very visible difference in responsiveness---it takes a fraction
of a second to react to clicks.  If it had sound effects, it would sound
like "clickCLUNK" instead of just "click".

(By the way, the four 2.9 images I tried were the ones from the 2.9alpha ftp
directory.  Adding the interCycle <= didn't affect responsiveness on 2.9;
didn't check 2.8.  higherPerformance didn't affect in two 2.9s or 3.0.a0)

I think I remember this response time change on my laptop, which is a
Pentium 233, but wrote it off to, well, test piloting and not having the
right VM.  I bet it will be a lot easier to notice this (if it exists for
other people) on slower machines.

I'm hoping somebody will come along and point out that I've got some doodad
misconfigured, or my machine sucks---I'm quite willing to believe this is a
local problem.

Jay