Work out gustafson speedup instantly with clear inputs, formula shown and shareable results.
Gustafson's law assumes the problem grows with the machine, giving S = s + p x N — linear in processor count rather than asymptotically capped. It is the right model for weak scaling, where a bigger machine is used to solve a bigger problem in the same time.
Gustafson's law
S = s + (1 - s) x N, where s is the serial fraction of the scaled workload
When the workload scales with the machine — larger simulations, finer meshes, bigger models. Amdahl applies to fixed-size problems (strong scaling).
Gustafson, generally. Nobody buys a bigger cluster to run the same small problem faster; they run a larger problem.