BinProlog Naive Reverse Benchmark.
Results in KLIPS, i.e. thousands of inferences per second.
It seems that even on mono-processor NTs, there's some speed-up for multi-threaded
execution,
probaly due to improved locality of reference. For large values of the reversed
list's length, simultaneus
garbage collection may occur on each thread, independently.
Source
CGI library
NEW multi-threaded BinProlog