Fast, single-file, MIT-licensed large integer factorization using ECM combined with other techniques.
Absolutely demolishes Flint, GMP-ECM, and other ECM-based factorization libraries in performance (though if you need to factor HUGE numbers, better, more complicated techniques than ECM exist).
NO dependencies at all except GNU MP (a.k.a GMP, a.k.a. MPIR on Windows), the standard Big Integer library.
Linux/Windows both supported - libgmp on Linux, mpir.lib on Windows.
Be sure to compile GMP/MPIR configured specific to your architecture. The generic version is often more than 2.5x slower.
Simple call:
std::vector<FactorInfo> Factorize(mpz_srcptr n);You are returned a vector of FactorInfo structs, each containing an mpz_t (the factor) and a uint64_t (the exponent).
Returning from the function transfers ownership of the mpz_ts within. As the caller, YOU are responsible for mpz_clearing them when you're done with them.
Multi-threaded by default, with automatic selection of thread count. You can override with a custom thread count, including disabling MT by requesting just 1 thread, using a second argument:
std::vector<FactorInfo> Factorize(mpz_srcptr n, uint32_t numThreads = 0);0 is the default and means automatic selection of thread count.
For machine-word-sized (32- or 64-bit) integers, see https://github.com/komrad36/Factorization-Primality.