Automatic Design Exploration Framework for Multicores with Reconfigurable Accelerators
نویسندگان
چکیده
Despite their promising improvements in performance and power efficiency, the possibilities of reconfigurable computing in multiprocessor environments are still mainly unexplored. This paper presents a new framework for rapid testing of multicores with application-specific reconfigurable accelerators. The design exploration is a multi-step flow that performs automatic generation of fine-grained accelerators, automatic parallelization, and testing. The automatic generation of accelerators is accomplished by an in-house developed software. Automatic parallelization is performed by OSCAR (Optimally SCheduled Advanced multiprocessoR) source-to-source compiler. The testing platform is configured from 1 up to 8 cores, and each core has a private reconfigurable space to implement application-specific accelerators. We test the design process with applications from the media domain (Optical Flow, AAC encoder, MPEG2 encoder and MPEG2 decoder). The results show that the multicore configuration outputs of our design exploration framework for 8 CPUs with reconfigurable accelerators achieve a maximum speedup of 6.57x for the Optical Flow application.
منابع مشابه
Automatic Generation and Testing of Application Specific Hardware Accelerators on a New Reconfigurable OpenSPARC Platform
Specific hardware customization for scientific applications has shown a big potential to address the current holy grail in computer architecture: reducing power consumption while increasing performance. In particular, the automatic generation of domain-specific accelerators for GeneralPurpose Processors (GPPs) is an active field of research to the point that different leading hardware design co...
متن کاملPerformance Estimation for the Exploration of CPU-Accelerator Architectures
In this paper we present an approach for studying the design space when interfacing reconfigurable accelerators with a CPU. For this purpose we introduce a framework based on the LLVM infrastructure that performs hardware/software partitioning with runtime estimation utilizing profiling information and code analysis. We apply it to reconfigurable accelerators that are controlled by a CPU via a ...
متن کاملA Two-Level Hardware/Software Co-Design Framework for Automatic Accelerator Generation
This paper presents a novel hardware/software Co-Design framework CoDe-X for automatic generation of Xputer based accelerators. CoDe-X accepts C-programs and carries out both, the host/accelerator partitioning for performance optimization, and (second level) the parameter-driven sequential/structural partitioning of the accelerator source code to optimize the utilization of its reconfigurable d...
متن کاملAutomatic hardware generation for the Molen reconfigurable architecture: a G721 case study
The advantages of the reconfigurable technology in terms of performance have been widely recognized. However, programming reconfigurable systems and designing hardware accelerators for them is not a trivial task. The Molen paradigm provides an easy to use approach to couple a General Purpose Processor (GPP) with custom designed reconfigurable accelerators both at program level and at hardware d...
متن کاملMAMPSx: A Design Methodology for Rapid System-Level Exploration, Synthesis of Heterogeneous SoC on FPGA
To achieve better performance and to meet time-to-market demands, heterogeneous reconfigurable MPSoCs like Xilinx Zynq platform are fast becoming popular. However, there are very limited design tools that enable both programming of applications and the exploration of the design space of these applications on the heterogeneous platforms. Also these existing tools are time consuming and require e...
متن کامل