skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: MATLAB-Based VHDL Development Environment

Conference ·
OSTI ID:976081

The Reconfigurable Computing program at Los Alamos National Laboratory (LANL) required synthesizable VHDL Fast Fourier Transform (FFT) designs that could be quickly implemented into FPGA-based high speed Digital Signal Processing architectures. Several different FFTs were needed for the different systems. As a result, the MATLAB-Based VHDL Development Environment was developed so that with a small amount of work and forethought, arbitrarily sized FFTs with different bit-width parameters could be produced quickly from one VHDL generating algorithm. The result is highly readable VHDL that can be modified quickly via the generating function to adapt to new algorithmic requirements. Several additional capabilities are integrated into the development environment. These capabilities include a bit-true parameterized mathematical model, fixed-point design validation, test vector generation, VHDL design verification, and chip resource use estimation. LANL needed the flexibility to build a wide variety of FFTs with a quick turn around time. It was important to have an effective way of trading off size, speed and precision. The FFTs also needed to be efficiently implemented into our existing FPGA-based architecture. Reconfigurable computing systems at LANL have been designed to accept two or four inputs on each clock. This allows the data processing rate to be reduced to a more manageable speed. This approach, however, limits us from using existing FFT cores. A MATLAB-Based VHDL Development Environment (MBVDE) was created in response to our FFT needs. MBVDE provides more flexibility than is available with VHDL. The technique allows new designs to be implemented and verified quickly. In addition, analysis tools are incorporated to evaluate trade-offs. MBVDE incorporates the performance of VHDL, the fast design time of core generation, and the benefit of not having to know VHDL available with C-tools into one environment. The MBVDE approach is not a comprehensive solution, but is a powerful method for algorithms that involve the cascading of fundamental building blocks.

Research Organization:
Los Alamos National Laboratory (LANL), Los Alamos, NM (United States)
Sponsoring Organization:
USDOE
OSTI ID:
976081
Report Number(s):
LA-UR-02-0852; LA-UR-02-852; TRN: US201009%%531
Resource Relation:
Conference: Submitted to: Engineering of Reconfigurable Systems and Algorithms, June 24-27, 2002, Las Vegas, Nevada
Country of Publication:
United States
Language:
English