3 ms·
Uses the wolfsort benchmark, which uses a time-based seed for the pseudo-random-number generator "rand". These are therefore unreproducible results (for the ra
by j_not_j 4y ago
Uses the wolfsort benchmark, which uses a time-based seed for the pseudo-random-number generator "rand". These are therefore unreproducible results (for the randomized inputs).
The results are very interesting and generally follow the rule that bigger hand-optimized code can be faster code.
- DennisP 4y agoMaybe not precisely reproducible, but they ran each test a hundred times so it should be possible to reproduce it well enough in statistical terms.
- tuukkah 4y agoIn the code it looks like the seed to the benchmark can be provided as the 4th command line argument if necessary: https://github.com/scandum/quadsort/blob/master/src/bench.c#L631 https://github.com/scandum/quadsort/blob/master/src/bench.c#...