| Summary: | Multi-core processors are everywhere now, researchers all over the world are finding
ways to benchmark scheduling algorithms to better make use of the multi-core
processing capability. However, researchers are still using only single core to
benchmark their algorithms despite having hardware with multiple cores resulting in
slow progress.
Code concurrency have been around for very long and there are many ways to
achieve code concurrency, threads and are two strategies of doing so. This report
explores into the possible combination of threads and processes namely, pure multithreading,
pure multi-core processing, nesting multi-threads in multi-core processes
and nested multi-core processes to achieve better speedup performance of
benchmarking algorithms. The performance of each strategy was measured and
compared, the best strategy is chosen to be optimized and used in creating a simulation
platform for researchers all around to use this paralleled processing environment
through the means of remote access and graphical user interface (GUI).
Nesting multiple processes yields the best performance among the strategies
and having the right amount of nesting is key to optimize the benchmarking of
individual algorithm. Having too much nesting would results in deterioration of the
parallelization performance.
|
|---|