Performance and Overhead Measurements
The previous sections used a set of parameters (overheads, number of processors, etc) to analyze and predict the performance of parallel programs. This section addresses the problem of measuring those and other important parameters, for a given shared memory machine. These parameters are not likely to be found in the manuals, since the combination of the operating system with the hardware is not usually documented (it depends for instance, on the underlying compiler). In this section we use a specific multicore parallel machine called MC, however the proposed methods should work for other machines as well.
KeywordsShared Variable Cache Line Local Access Cache Coherency Protocol Multicore Machine
- Eggers, S.J., Katz, R.H.: The effect of sharing on the cache and bus performance of parallel programs. In: Proceedings of the Third International Conference on Architectural Support for Programming Languages and Operating Systems, pp. 257–270. ACM, New York (1989). ISBN 0897913000 CrossRefGoogle Scholar
- Hristea, C., Lenoski, D., Keen, J.: Measuring memory hierarchy performance of cache-coherent multiprocessors using micro benchmarks. In: ACM/IEEE 1997 Conference Supercomputing, p. 45 (1997) Google Scholar
- Jost, G., Jin, H., an Mey, D., Hatay, F.F.: Comparing the openmp, mpi, and hybrid programming paradigms on an smp cluster. In: Proceedings of EWOMP, vol. 3 (2003) Google Scholar