I'll stick with CryptoNight vs Cuckoo Cycle, assuming a size 2^{32} for the latter
memory 2MB 4GB
memoy-hardness solid can run twice as slow in 3GB, or 50% faster in 6GB
hardness hypothesized; not provable
parallellizable not at least 20+ threads (probably less than 100)
access pattern random random
latency low (cache) high (main memory)
verification slow instant
%computation probably > 90% 33%
%memory probably < 10% 67%
Can you explain what latency means in this particular case? I mean I know it's some kind of delay between something but what stands for "something" here?
Latency is the time between the start of a memory access instruction, and it's end,
when the value has been retrieved, and is available for use.
And from your numbers I assume that cache is much faster compared to main memory?