So, the results of testing Argon2d.
Initially, the idea was to take a target in 1G RAM to protect against botnets.
However, in real mining with such a goal, 1 hash is calculated for a few seconds and the system falls into a coma.
Therefore, smaller targets were tested - from 64MB and down twice at each iteration.
The results are somewhat confusing, but the conclusions are as follows:
- to protect against botnets Argon will not work, because RAM consumption is not too large;
- on large targets, powerful processors do not exceed low-power ones;
- CPU load increases linearly with the increase in the number of threads and does not depend on tagret, except for tiny ones;
- on tiny targets powerful processors begin to dominate.
Please express your thoughts, what size of target we will take in PoW ...