Why do the miller_rabin_test() at all?

[tavianator]

It should be sufficient, and much faster, to just check if (P > 1 && N % P == 0) right?

[leonklingele]

👍 but you need to check for P != N as well.

[leonklingele]

Throwing out normalizedEntropy will then make it even faster.

[aguinet]

That's stuff that are there because I was checking this step by step. I haven't done yet an optimization pass phase, so you might be right, but real-life benchmarks will tell! (need to do them!)

[tavianator]

@leonklingele

but you need to check for P != N as well.

In general yeah. But if P is always 1024 bits and N is always 2048 bits then you don't need to :)

[aguinet]

So I did some benchmark and you're right, the miller rabin test was really slowing down the whole thing. Fixed code and benchmarking code has just been pushed!