2 ms·
Very efficient for storage and memory bandwidth, but such a scheme is a headache for high throughput hardware implementations (at least compared to regular 4 bi
by londons_explore 3y ago
Very efficient for storage and memory bandwidth, but such a scheme is a headache for high throughput hardware implementations (at least compared to regular 4 bit math, which can be packed really really densely)