@qiemem reported the Rock Paper Scissors library model performed poorly in NetLogo Web. In checking, it was an issue with the one-of primitive being used in combination with large agent sets, specifically a large world patch set. The improvements should help speed up any model that uses one-of with agent sets, and especially so with larger agent-sets.
These changes have been reviewed by @TheBizzle already, I just wanted a place to post the optimization results for RPS. Qualitatively the model is also much more usable with the changes. Other model benchmarks showed no change, or more modest improvements.
Time: 08/09/2018 @ 12:40:00 PM CDT
Version: e50ad8c-dirty
Models: Rock Paper Scissors
Iterations: 20
Ticks: 200
Engines: V8
Comment: `one-of` pre-optimizations
Rock Paper Scissors (V8 Node.js v9.7.1):
--Average: 446.356 seconds
--Min: 439.326 seconds
--Max: 452.935 seconds
==========
Time: 08/09/2018 @ 12:22:52 PM CDT
Version: 8f627a5-dirty
Models: Rock Paper Scissors
Iterations: 20
Ticks: 200
Engines: V8
Comment: `one-of` post-optimizations
Rock Paper Scissors (V8 Node.js v9.7.1):
--Average: 26.993 seconds
--Min: 26.285 seconds
--Max: 27.482 seconds
@qiemem reported the Rock Paper Scissors library model performed poorly in NetLogo Web. In checking, it was an issue with the
one-ofprimitive being used in combination with large agent sets, specifically a large world patch set. The improvements should help speed up any model that usesone-ofwith agent sets, and especially so with larger agent-sets.These changes have been reviewed by @TheBizzle already, I just wanted a place to post the optimization results for RPS. Qualitatively the model is also much more usable with the changes. Other model benchmarks showed no change, or more modest improvements.