Parallel loop scheduling using an evolutionary algorithm

Using of Cuckoo Search Algorithm and Wave-Fronts Approach with Changing Angle for Tiled Iteration Space Scheduling of Two-Level Nested Loops

“…According to our related approaches [8][9][10][11], we run the simulated software for various experiment and select some of them here. …”

“…Fig. 3: An example of wave-front approach and it's Angle [9] As shown in the above figure -part a, there are one tile on the first and last waves, two tiles on the second and penultimate waves, and maximum number of tiles on the middle wave. But part b shows that the number of existing tiles on each wave is almost equal.…”

“…In some methods, in order to minimize the overall execution time of iterations, the communication time of processors overlaps with the time of internal calculations of processors [3, 7 and 24]. Also some other approaches inspired by the wave-front method, eliminate the need for communicating of many processors and assign dependent caches to the same processor [7,9,15,24]. In other approaches, the critical path for the task graph resulted from the loop iterations' space is calculated and in order to obtain load balancing, other non-critical iterations are distributed among processors according to their data dependency [1].…”

“…1: An example of nested loop, it's iteration space and tiled iteration space [9] The command inside the loops indicates that the iteration […”

Design methodology for workload‐aware loop scheduling strategies based on genetic algorithm and simulation

“…Their results pointed out that a hybrid scheduling strategy may lead to up to 4.51 of performance speedup over OpenMP scheduling solutions. Another work that also relies on compiling information is presented in .…”