ReOPN: Retargetable Statistical Optimisation and Parallelisation for Network Processors

Project Details

Key findings

We have developed a methodology for the characterisation of computer workloads, which are subsequently visualised for human inspection and interpretation. We have found that this workload characterisation approach is very effective in guiding compiler developers to pinpoint performance bottlenecks and to devise new compiler optimisations. This workload characterisation has been applied to networking applications, where we have found that data layout transformations are key to enabling higher performance due to data cache contention. Finally, we have developed new compiler-based data transformations that for a class of data packet processing applications can improve performance by up to a factor of 2.
Effective start/end date21/09/0720/09/10


  • EPSRC: £142,784.00


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.