Closed kovacs-levent closed 3 years ago
I've implemented a divide and conquer recursive approach with adding in parent-threading by hashing parent nodes on multiple threads if possible.
It's now faster than the single-threaded version, but I think we need SIMD to really be able to be efficient with the multi-threading.
Investigate how it's possible to parallelize Python programs in general and whether we can parallelize our hash function.
This is closely tied to performance testing, if parallelization is possible then our performance will increase. Based on the literature, BLAKE3 should provide the possibility for this.