Get the latest tech news
Optimizing a bignum library for fun
Can I make math operations run 100x faster on really, really big numbers?
There are tradeoffs for how many bits to use per chunk: memory, arithmetic performance, cache locality, overflow management, native CPU operations, complexity of our code, etc. Alright, so we have made two major optimizations to the bignum library and measured the impact on running time. Negative numbers Subtraction and division Exponentiation Bitwise operations Converting to/from other types Tests(I bet there are many bugs in my code)
Or read this on Hacker News
