@alcinnz why is it typically the LSBs?
Post
Replies:
6
@alcinnz why is it typically the LSBs?
@hipsterelectron Because they represent superhuman precision, beyond what we're consciously controlling.
Our scheduler would supply the exact timing of interrupts, & drivers could supply additional randomness. Also we could add dedicated hardware. Another hash function could supply programs with this randomness without exposing potentially-exploitable state.
Linux uses ChaCha20 & Blake2s for this, & I'd do likewise.
They'd run in the same coprocessor mentioned yesterday for counting the time, in which I'd include a barrel shifter to aid hashing & number parsing.
2/2! Tomorrow: General I/O!
@alcinnz i plan to implement blake3 soon because i'm very annoyed with the reference implementation and told them so
