17 Mar
2010
17 Mar
'10
4:35 a.m.
I should also note that these timings are quite temperamental on my computer; using profiling or removing the performGC calls can make C faster than D and B faster than A.
Binomial priority queue timings (in milliseconds, averaged over 20 runs) A: Simple Brodal-Okasaki list implementation: 199.3697 B: Wasserman nested type implementation, simple functions: 235.0143 C: Wasserman's 'Quick' list implementation: 179.7227 D: Wasserman's proposed nested type implementation: 163.47515
D is still the fastest. However, A is a good bit faster than B. Is it possible that C would show a similar advantage over D if it were optimized as carefully as D?