4 ms·
I meant run on the cpu itself. Most processors have floating-point support, and fixed-point just uses integer operations. All those heavy libraries require an a
by the_gorilla 2y ago
I meant run on the cpu itself. Most processors have floating-point support, and fixed-point just uses integer operations. All those heavy libraries require an arbitrary amount of memory for their calculations.
- dzaima 2y agoThe article here is about hardware without native FP already; that said, rationals (even before getting to the need of bigints) are quite bad regardless, likely requiring division & gcd around basic arith, which are slow both in software and in hardware.