Closed d-led closed 5 years ago
Benchmark shows, it's a space vs. speed trade-off, as expected
Benchmark mean median deviation iterations
Bitarray Push 10807 ns 10812 ns ±0.88% 10000
BoolArray Push 2360 ns 2357 ns ±1.12% 100000
Bitarray Apply 5224 ns 5244 ns ±1.80% 30000
BoolArray Apply 520 ns 506 ns ±7.76% 300000
this probably won't be needed at the moment. Ported the bit array library to todays's Pony: https://github.com/d-led/pony-bitarray, with which the benchmark has been produced
the current Dense Matrix is based on a vector of Bools. This might be a waste of space.
TODO: analyze the space taken by the bool vector, and create a BitArrayDenseMatrix, e.g. based on lisael/pony-bitarray (fixed version) to see and benchmark #4, whether this is valuable