DDM-VMc: the data-driven multithreading virtual machine for the cell processor

Abstract

In this paper we present the Data-Driven Multithreading Virtual Machine for the Cell Processor (DDM-VMc). Data-Driven Multithreading is a non-blocking multithreading model that decouples the synchronization from the computation portions of a program allowing them to execute asynchronously in a data-flow manner. The core of the DDM model is the Thread Scheduling Unit (TSU), which schedules threads dynamically at runtime based on data availability. DDM-VMc implements the TSU as a software module running on the PPE core of the Cell, allowing the SPE cores to execute the program threads. DDM-VMc virtualizes the parallel resources of the Cell, handles the heterogeneity of the cores and manages the Cell memory hierarchy efficiently. We present the architecture of DDM-VMc and provide an in-depth performance analysis using a suite of standard computational benchmarks. The evaluation shows that DDM-VMc scales well and tolerates scheduling overheads and memory latencies effectively. Furthermore, DDM-VMc compares favorably with other platforms targeting the Cell processor.