half_float
half_float copied to clipboard
Speed of half_float
Hey couple of questions,
- Is this faster than float arithmetic? Have you run some benchmarks?
- How fast or slow would this be compared to just casting to float and using uint16_t for moving data around.
This will be magnitudes slower than float arithmetic. It's really intended for cases where you'd want to convert float or double data to 16 bit floats for use in GPU buffers. The arithmetic operations that are implemented were mostly an exercise ;-)