Evaluation of Work Stealing Algorithms
This repository contains a Benchmark for the evaluations of different work stealing algorithms. Currently, the system can supports
- FIFO
- LIFO
- Priorityes (developped for this work)
For each algorithm we execute 30 Iterations with a varying number of cores (1..96). Our algorithm is tested on generated task DAGs of sizes 50..1600 with a varying density in the set {0.2, 0.5, 0.8} for each iteration and number of cores combination. The results of benchmarks are summarized at our online webpage