A Performance Prediction for Automatic Placement of Heterogeneous Workloads on Many-cores

Conference of 9th IEEE International Symposium on Embedded Multicore/Manycore SoCs, MCSoC 2015 ; Conference Date: 23 September 2015 Through 25 September 2015; Conference Code:118611International audienceCurrent trends in computer architecture show that we are aiming toward more cores and even more so more heterogeneity. As an extensive knowledge of processor's internals cannot be a prerequisite to their programming and for the sake of portability, these systems necessitate the compilation flow to evolve and cope with heterogeneity issues. In this paper, we lay a first step toward a possible solution to this challenge by exploring the results of Single Program Multiple Data (SPMD) type of parallelism with heterogeneous compute kernels and predicting performance of the compilation results so that our tools can guide a compiler to build an optimal partition of task automatically, even on heterogeneous targets. We explore a synchronous execution model and use execution time from application parts to predict the performance of the whole application. We show on experimental results a very good accuracy of our tools to predict real world performance on 3 case studies with current days server processors used as proxy experimental setup for future embedded manycores