The output of the test script from the other day.
| test ord | test := IdentitySet new. [ test size >= 1000000 ] whileFalse: [ ord := OrderedCollection new: 100. 10000 timesRepeat: [ test add: ( ord add: Object new ) ]. Transcript show: test size asString; tab. Transcript show: [ 10 timesRepeat: [ ord do: [ :each | test includes: each ] ] ] timeToRun asString. Transcript cr ]
On Mon, Mar 22, 2010 at 4:46 PM, Bert Freudenberg bert@freudenbergs.de wrote:
On 22.03.2010, at 22:40, Chris Muller wrote:
4.1 hashed collections, across the board, small to large, are slower by a factor of 2?! I just don't think we can keep doing this; getting slower and slower and slower, like molasses.. I'm sorry, but I really care about this and I know you do too because speed was the whole premise of putting these changes in.
What went wrong? More importantly, how can we fix this? <IdentitySet.3.9.png><IdentitySet.4.1.png><IdentityDictionary.3.9.png><IdentityDictionary.4.1.png>
Neat graphs. But what exactly are you measuring?
- Bert -