A Virtual SIMD Machine Approach for Abstracting Heterogeneous Multicore Processors
نویسندگان
چکیده
The heterogeneous design of multi-core processors, such as the Cell processor, introduced new challenges in porting high-level languages. Our project is developing tools that hide the underlying details of the Cell processor and eases parallel programming. We present a Virtual SIMD machine (VSM) paradigm that can be used to parallelize array expression automatically. The novelty is the use of a virtual SIMD machine model to completely hide the underlying details required for programming the Cell processor. The VSM paradigm can also be used to develop an automatic parallelizing compiler for the Cell Broadband Engine (Cell BE). In this paper we give an overview of the VSM interface and present preliminary results that show the performance of our VSM and its behavior on multiple accelerator cores using basic arrays operations. Keywords-component; High-level Languages, Virtual Machine, Parallel techniques, Multicore Compiler.
منابع مشابه
Hera-JVM: Abstracting Processor Heterogeneity Behind a Virtual Machine
Heterogeneous multi-core processors, such as the Cell processor, can deliver exceptional performance, however, they are notoriously difficult to program effectively. We present Hera-JVM, a runtime system which hides a processor’s heterogeneity behind a homogeneous virtual machine interface. Preliminary results of three benchmarks running under Hera-JVM are presented. These results suggest a set...
متن کاملHeterogeneous Chip Multiprocessor Design for Virtual Machines
Multicore architectures provide an alternative to increasing clock frequencies to improve performance of modern processors. The best design for these chip multiprocessors, including structure sizes or whether to use homogeneous or heterogeneous cores, remains open for exploration. One potential design path involves using heterogeneous cores that are specialized for a given task. This paper exam...
متن کاملChapter 8 A Programming Environment forNetwork
Heterogeneous associative computing (HAsC) is a new network heterogeneous computing (NHC) paradigm that is a combination of associative computing and heterogeneous computing as related to superconcurrency. The goal of this high performance computing environment is to increase the throughput of very large applications such as those on the scale of grand challenge problems. HAsC is the result of ...
متن کاملSupporting Transparent Thread Assignment in Heterogeneous Multicore Processors Using Reinforcement Learning
Heterogeneity in multicore processor systems creates challenges in effectively mapping processes to diverse cores. While most approaches require programmer partitioning between core types or permutation of thread schedules to find the optimal mapping, we introduce a new machine learning approach to automated thread assignment. We train a reinforcement learning agent to assign threads to the bes...
متن کاملHigh-Efficient Parallel CAVLC Encoders on Heterogeneous Multicore Architectures
This article presents two high-efficient parallel realizations of the context-based adaptive variable length coding (CAVLC) based on heterogeneous multicore processors. By optimizing the architecture of the CAVLC encoder, three kinds of dependences are eliminated or weaken, including the context-based data dependence, the memory accessing dependence and the control dependence. The CAVLC pipelin...
متن کامل