Morphic slowness (was Re: Does *anyone* use MVC?)
Bob Arning
arning at charm.net
Fri Aug 2 09:38:42 UTC 2002
On Thu, 1 Aug 2002 22:26:00 -0700 Dan Ingalls <Dan at SqueakLand.org> wrote:
>So from my brief exploration, it seems that as yet unknown factors have slowed browser creation by almost 100%, and then we've added another 50% in frills that can be turned off.
I ran this test in 3.2 vs. 2.8 and got about 9 seconds vs. 6 - or about 50% slower.
Some of the factors in the difference:
-----------------------------------------
2.8:
| 9.6% {587ms} PluggableTextMorph class>>on:text:accept:readSelection:menu:
3.2:
| | 17.7% {1637ms} PluggableTextMorph class>>on:text:accept:readSelection:menu:
and a big chunk of that is due to...
| | | 14.9% {1378ms} Browser>>selectedMessage
| | | 8.9% {823ms} EventSensor(InputSensor)>>controlKeyPressed
| | | |8.9% {823ms} EventSensor>>primMouseButtons
which doesn't appear on the radar in 2.8
-----------------------------------------
in 2.8
| | 5.1% {312ms} SystemWindow(Morph)>>bounds:
| | 5.1% {312ms} SystemWindow>>extent:
| | 5.1% {312ms} SystemWindow>>setBoundsOfPaneMorphs
| | 5.1% {312ms} PluggableListMorph(Morph)>>bounds:
| | 2.2% {135ms} PluggableListMorph(ScrollPane)>>extent:
| | 2.2% {135ms} PluggableTextMorph>>extent:
in 3.2:
| | 13.1% {1211ms} SystemWindow(Morph)>>bounds:
| | 13.1% {1211ms} SystemWindow>>position:
| | 13.1% {1211ms} SystemWindow(Morph)>>position:
| | 13.1% {1211ms} SystemWindow(Morph)>>fullBounds
| | 13.1% {1211ms} SystemWindow(Morph)>>doLayoutIn:
| | 11.5% {1063ms} ProportionalLayout>>layout:in:
| | 11.5% {1063ms} AlignmentMorph(Morph)>>layoutProportionallyIn:
| | 9.1% {841ms} AlignmentMorph(Morph)>>fullBounds
| | |9.1% {841ms} AlignmentMorph(Morph)>>doLayoutIn:
| | | 8.9% {823ms} ProportionalLayout>>layout:in:
| | | 8.9% {823ms} PluggableTextMorph(Morph)>>layoutProportionallyIn:
| | | 8.3% {767ms} PluggableTextMorph(Morph)>>layoutInBounds:
| | | 8.3% {767ms} PluggableTextMorph(Morph)>>bounds:
... much truncated...
so we pay a bit for a more flexible layout scheme
-----------------------------------------
finally, although the 2.8 MessageTally did not report GC's, I did notice that my 3.2 test lost 21% to GC:
full 1 totalling 601ms (7.0% uptime), avg 601.0ms
incr 728 totalling 1,285ms (14.0% uptime), avg 2.0ms
-----------------------------------------
Cheers,
Bob
More information about the Squeak-dev
mailing list
|