When scheduling tasks on a processor core, why does simply summing task durations incorrectly estimate total runtime if some tasks access shared memory resources?

💡 Explanation

Summing task durations overcounts the true runtime because the mutual exclusion mechanism of critical sections enforces serialization on shared resources, therefore actual runtime includes waiting, rather than only execution, unlike the isolated case. Cache misses or branch prediction issues are performance factors but do not directly invalidate simple summation.

• A municipal water distribution system experiences fluctuating demand; which effect results when the system of linear equations representing water flow and pressure at each node becomes underdetermined?
• If a social network mandates that among any six users, either three are mutually friends or three are mutually strangers, which consequence follows for the network's structure?
• If a university course timetable can be represented as a bipartite graph, where courses and time slots are nodes, which outcome guarantees conflict-free scheduling?
• A wireless router allocates bandwidth to connected devices using a greedy algorithm; which consequence follows?
• If a matrix representing the connectivity of a sensor network becomes singular, which outcome is most likely?
• A propositional logic solver using the DPLL algorithm encounters a CNF formula with a unit clause. Which consequence follows during unit propagation?