Benchmarking Multi-Core Processors

From Master Projects
Jump to: navigation, search


About Benchmarking Multi-Core Processors

  • This project has not yet been fulfilled.
  • This project fits in the following Bachelor programs: {{#arraymap:|, |xXx|bachelorproject within::xXx|,}}
  • This project fits in the following masterareas: {{#arraymap:High Performance Distributed Computing, Parallel and Distributed Computer Systems|, |xXx|project within::xXx|,}}


Description

In the early 2000s, increasing the frequency of single-core processors has reached the point of diminishing returns. Multi-core processors have emerged as an alternative approach for increasing processor performance: instead of having a single, large, very complex, power hungry core per processor, these new computing architectures combine multiple "simplified" cores, which, by running concurrently, achieve better overall performance than the equivalent single-cores.

Several types of multi-core architectures have emerged, making classifications and comparisons between them non-trivial. This project focuses on developing a strategy and a toolkit for benchmarking multi-core processors. The proposed strategy has to include a set of metrics, as well as means to measure them. The toolkit should include a set of representative applications (synthetic and/or complex) that focus on the proposed metrics and can quantify them in a deterministic way; the applications should be implemented in a common language (e.g., OpenCL), able to accommodate all studied architectures.