[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]
lite MOOSE measurement
- To: <xtech@xxxxxxxxxxx>
- Subject: lite MOOSE measurement
- From: Paul Baclaski <autodesk!thezahir!peb>
- Date: Wed, 19 Dec 90 17:23:37 PST
In one test of the liteness of the new MOOSE, I compared the old
genetic Barnsley fractal genetic system of 10 individuals.
Heap size excluding 34k base and 28k of class objects:
Old: 422k
New: 149k
Approximately 350 instance objects are created, with an average savingsof 780 bytes per instance object.
The lite objects (instances) now use up 430 bytes instead of 1210 bytes.
This 430 bytes appears to be composed of 30 cons cells (120 bytes),
plus bindings (and lexical scoping overhead, how ever that is
represented).
[I used these numbers to estimate the space requirements for
100 individuals and the estimate was off by 2%, so these numbers
should be fairly accuate.]
Paul