Latest News/Update

We are analyzing portability of OpenCL code on a variety of platforms, specifically relating to floating point behavior. This project is funded by NSF grant CCF-1218075 Ensuring Reliability and Portability of Scientific Software for Heterogeneous Architectures. We present new results. Floating Point Arithmetic for Heterogeneous Architectures

A new version of the VFLOAT variable precision floating point library is now available! The library has been reorganized for ease of use. All components have been tested up to double precision. All components work on Altera and Xilinx platforms.

Prof. Leeser is featured on NVIDIA's Women and CUDA blog on the second page.

A paper titled: "Fast Reconstruction of 3D volumes from 2D CT projection data with GPUs" has been published by Biomed Central Research Notes. More information is available at our 3D CT Reconstruction web page.

Jonathon Pendlum successfully defends his MS thesis, CRASH: Cognitive Radio Accelerated with Software and Hardware, April 2014. More details on his research project CRASH are available here, including software for download. A paper on CRASH was presented at FCCM 2014 and a talk at NEWSDR.

Field Programmable Gate Array

Field Programmable Gate

Instructions for building one :)

Acceleration of Algorithms using FPGAs and GPUs

Research in the Reconfigurable and GPU Computing Laboratory (RCL) investigates the use of Accelerators such as Field Programmable Gate Arrays (FPGAs) and Graphics Processing Units (GPUs) for scientific applications. Our interest is in developing libraries and interfaces that make acceleration easier to use. In GPU research, we are investigating how best to take advantage of GPUs from common programming environments such as Matlab and SCIRun. We are also looking at Tasks and Conduits as a mechanism for mapping the same application to several different target hardware platforms.

Our FPGA projects include using OpenCPI as a component model for FPGA research. In addition, we have developed a widely used variable precision floating point library (VFLOAT). In the lab we use a combination of research and commercial tools, as well as the software and hardware needed to map designs onto field programmable logic.

Reconfigurable and GPU Computing Lab Links

<<   Photos   >>

Funding and Donations

Research Funding in The RCL comes from:

We would like to thank the companies who donate hardware and software to the RCL: