Fork me on GitHub

Had some discussions about performance for our use cases, and I made a bunch of assumptions. 1) In terms of inserts/retracts and fire-rules, Clara is single-threaded and does not take advantage of parallelziation. 2) Working set size and memory bandwidth dominate CPU speed.


Do those sound accurate?


@eraserhd (1) seems accurate enough; I think (2) isn’t clear enough to me to say yes/no


The CPU is typically dominated by the number of constraints being evaluated against the working set of facts. So yes to that part I believe. What do you mean by memory bandwidth?