1 | This will only depend on 3 parameters |
2 | n which is grain size -- amount of problem stored on each processor (bounded by local memory) |
3 | tfloat which is typical time to do one calculation on one node |
4 | tcomm which is typical time to communicate one word between two nodes |
5 | Most importance omission here is communication latency |
6 | Time to communicate = tlatency+ (Num Words)tcomm |
7 | Node A |
8 | Node B |
9 | tcomm |
10 | CPU tfloat |
11 | CPU tfloat |
12 | Memory n |
13 | Memory n |