3 ms·
Given so many fine gradiations of slice-size, it'd be interesting to see an analysis, for distributable workloads, when it's better to use more smaller slices v
by gojomo 17y ago
Given so many fine gradiations of slice-size, it'd be interesting to see an analysis, for distributable workloads, when it's better to use more smaller slices vs. fewer larger slices of similar cost. (On what sort of jobs will 12 256slices for $240 beat one 4096 slice for $250, and so forth?)