Scalability bottlenecks discovery in MPSoC platforms using data mining on simulation traces

• DOI: 10.7873/date2014.199
• Publication Date: 2015-04-03T12:53:14Z
• Authors: Sofiane Lagraa; Alexandre Termier; Frederic Petrot
• Keywords: MPSoC; Profiling (computer programming); Multi-core processor; Code (set theory)

Nowadays, a challenge faced by many developers is the profiling of parallel applications so that they can scale over more and cores. This especially critical for embedded systems powered Multi-Processor System-on-Chip (MPSoC), where ever demanding have to run smoothly on numerous cores, each with modest power budget. The reasons lack scalability are numerous, it be time consuming developer pinpoint correct one. In this paper, we propose fully automatic method which detects instructions code lead scalability. based data mining techniques exploiting low level execution traces produced MPSoC simulators. Our experiments show accuracy proposed technique five different kinds applications, how information reported exploited application developers.