Two Simulation Frameworks, M5sim and GEMS, Merge into GEM5.
Posted on 16-5-2011 by Zhenyu Ye
Tags: simulator, architecture
The GEM5 merges the M5sim simulator and the GEMS simulator. It seems the GEM5 no longer depends on the Virtutech Simics.
Related Works in Field-Programmable Custom Computing Machines (FCCM) 2011
Posted on 12-5-2011 by Zhenyu Ye
Tags: FPGA, CUDA, architecture, compiler, OpenCL
There are two papers in FCCM 2011 on frameworks supporting CUDA-like and OpenCL-like languages on FPGAs:
- Multilevel Granularity Parallelism Synthesis on FPGAs. It extends the FCUDA framework to support more levels of parallelism to compile high level language (CUDA-like) to FPGAs.
- Synthesis of Platform Architectures from OpenCL Programs. Its framework, named SOpenCL, uses similar approach like the FCUDA and the OpenRCL framework. SOpenCL differs from FCUDA in that SOpenCL has an architecture template that can be instantiated during the hardware generation, while FCUDA relies only on high level synthesis tool to generate the hardware. SOpenCL differs from OpenRCL in that SOpenCL’s architecture template is SIMD-like cores with customizable data path, while OpenRCL uses MIMD-like architecture template.
PetaBricks: A Language and Compiler for Algorithmic Choice
Posted on 10-5-2011 by Zhenyu Ye
Tags: architecture, compiler
PetaBricks, a project leaded by the compiler group of MIT, has recently released the source code of this tool.
Posted on 2-5-2011 by Cedric Nugteren
The UK-based company Clearspeed has a nice parallel architecture with the CSX700. Go to their website for more details. It runs at 250MHz, uses 9W and can do 96GFLOPS single or double precision.
There’s a low-profile PCI-Express board available (here on an Intel Atom board):
IEEE Micro Special Issue on Hot Chips 2010
Posted on 2-5-2011 by Zhenyu Ye
Tags: architecture, GPU, conference, multicore, NOC
IEEE Micro March/April 2011 is a special issue on Hot Chips 2010.
Dynamically Reconfigurable Systems Architectures, Design Methods and Applications
Posted on 26-4-2011 by Zhenyu Ye
Dynamically Reconfigurable Systems Architectures, Design Methods and Applications, edited by Marco Platzner, Jürgen Teich and Norbert Wehn, a book published in 2010. The content is available from the TU/e digital library.
A Special Issue: IBM POWER7 Technology and Systems
Posted on 21-4-2011 by Zhenyu Ye
The IBM Journal of Research and Development has a special issue on IBM POWER7 Technology and Systems. “The papers in this issue describe the key hardware and software technologies used to build POWER7 processor-based systems from chip manufacturing technologies and architectural design and verification to systems packaging, design, and performance.”.
HiPEAC internships 2011
Posted on 20-4-2011 by Cedric Nugteren
Tags: algorithmic skeletons, internships
As a reminder, the HiPEAC internships are still available through the HiPEAC website.
Also at HiPEAC, a PhD position is open to work on C++ Metaprogramming and algorithmic skeleton for automatic code deployment and benchmarking on parallel embeded systems. It seems algorithmic skeletons are becoming popular again.