Aaron J Reichow <reic0024 at d.umn.edu> wrote:

> While I know that tinyBenchmarks aren't representative of total
> performance, I still think this is a bit odd.   Am I just over-simplifying
> things, or could there really be something here?
> 
> Are there problems or inefficiencies with the way Squeak does display or
> input events on the WinCE version of Squeak? The Axim is evidentally able
> to actually process data faster in Squeak- it just seems that Morphic-
> somehow- ends up being slower.
bytecodes/sec is almost purely a cpu speed/cache thing so it's not too
surprising that you get 30% or so faster on the Axim than the ipaq.
Sends/sec is likely to be a little more stressful on the memory system
if only because it does actually do a load of sends and thus exercises
methodcache lookup and context allocation/send/return. What's weird is
that it still gets less sends/sec than my tatty old Acorn, thus
demonstrating that a compiler that does actually undersand the cpu can
have a big effect.

Responsiveness in morphic has a lot to do with how well the event
input is done, the memory path from Display to glass and (I guess)
the more general memory bandwidth of the machine. Thus my NetWinder
(another ARM linux machine) is slower on tinyBenchmarks than my Acorn
but much more responsive in morphic.

tim
-- 
Tim Rowledge, tim at sumeru.stanford.edu, http://sumeru.stanford.edu/tim
Fractured Idiom:- IDIOS AMIGOS - We're wild and crazy guys!