Closed cisaacson closed 4 years ago
You should always benchmark things like this (preferably with something like criterion). It probably depends only how sparse the bitsets are and how you then eventually consume the information which bits are active.
Please do post your benchmark results if you create any :D
Thanks, I will perform some benchmarks once we accumulate more data. It's an interesting problem, the bitset
s will start out very sparse and then as we progress in processing they will be more dense, potentially up to full density. However, most of the processing is on the sparse bitset
s, so that's why I think hibitset
will win. I'll post something once I have conclusive results.
I have a use case with 100K bits, where I need to so simple AND/OR operations. It would seem based on the layer design that
hibitset
would be faster thanfixedbitset
.Any advice on this?