Catalytic Inc. Unveils Product for Streamlining the Implementation Of Complex Signal Processing Algorithms; Tool Speeds Simulation of Complex MATLAB Algorithms Used in Communication, Video ASIC Design.

PALO ALTO, Calif. -- Catalytic Inc., a company simplifying the signal processing design flow, today began shipping Catalytic RMS (Rapid MATLAB(R) Simulator), software for streamlining the algorithm design of signal processing systems eventually implemented in register transfer level (RTL) code.

Catalytic products shorten the design time for signal processing applications by providing fast simulation throughout the design flow and by greatly simplifying the floating- to fixed-point conversion process. Catalytic RMS is tailored for algorithms targeted to an RTL-based implementation. Catalytic has been shipping it first product, DeltaFX, since November 2004. DeltaFX is used by software engineers developing algorithms in the MATLAB environment targeted for digital signal processors (DSPs).

"With the introduction of Catalytic RMS, Catalytic is greatly assisting signal processing design groups who implement their algorithms in RTL," says Randy Allen, Catalytic's founder and chief executive officer (CEO). "As an example, engineers creating wireless ASICs describe their entire chip in MATLAB because it is an excellent language for describing the highly mathematical algorithms that comprise the core intellectual property of wireless systems. But the algorithm engineers must quickly move out of MATLAB because of the slow simulation speed and the lack of fixed-point modeling and analysis capability. Catalytic RMS can simulate core wireless algorithms 200 times faster than existing methods. MATLAB users no longer have to spend time hand translating MATLAB to C to achieve acceptable simulation speeds."

Introducing Catalytic RMS

Catalytic RMS works seamlessly with MATLAB (a popular numerical analysis and visualization environment from The MathWorks) to provide fast floating-point simulation for standard MATLAB code, accurate fixed-point modeling capabilities, and fast fixed-point simulation. The result is a streamlined algorithm design process that accelerates the simulation of standard MATLAB and assists with the conversion of floating-point MATLAB to optimized fixed-point MATLAB, which accurately models the behavior of the final RTL design.

Catalytic RMS is controlled through a feature-rich and customizable graphical user interface (GUI). Algorithms are loaded and can be immediately accelerated. The GUI also gives guidance on how to improve simulation execution speed to achieve maximum performance. When used to convert floating-point algorithms to fixed, it provides static code analysis to show that all variables have been converted to the appropriate fixed-point type. The GUI's value profiling capability helps the user choose the best fixed-point representations based on simulation runs and analysis.

Catalytic RMS speeds floating- and fixed-point MATLAB simulations using Catalytic's proprietary compiler technology, which generates a compiled library from MATLAB code. This library executes much faster than the original interpreted MATLAB code.

Fast Simulation with Catalytic RMS

Catalytic's RMS simulation engine allows developers to quickly accelerate MATLAB functions and use them throughout the design cycle. The simulation engine converts a MATLAB program into an accelerated library that executes much faster than the original MATLAB. It is integrated seamlessly with MATLAB, supports most MATLAB language constructs and the Signal Processing and Communications toolboxes from The Mathworks. Both floating-point and fixed-point algorithms can be accelerated providing developers the ability to convert designs to fixed-point while still working within the MATLAB environment.

In early customer benchmarks, Catalytic RMS typically accelerates the simulation execution of floating-point MATLAB by 10X to 50X. However, execution acceleration in excess of 200X is not uncommon. Fixed-point MATLAB simulation execution is typically accelerated by 100X to 1000X in these same benchmarks.

Pricing and Availability

Catalytic RMS runs on the Windows and Linux operating system. Catalytic RMS pricing starts at $15,000 for an annual license.

For more details, contact Lisa Schmidt, Catalytic's vice president of marketing. She can be reached at [phone] x 220 or via email at [e-mail]

About Catalytic Inc.

Catalytic speeds the development of signal processing applications by creating an efficient path from algorithmic specification to implementation. Catalytic Inc. was founded in 2001 by Dr. Randy Allen, a successful entrepreneur, world-renowned compiler technology expert and the co-author of the book, "Optimizing Compilers for Modern Architectures: A Dependence-based Approach." Corporate headquarters are located at 1076A East Meadow Circle, Palo Alto, Calif. 94303. Telephone: [phone]. Facsimile: [phone]. Email: [e-mail]. More information about Catalytic Inc. can be found at its website located at: http://www.catalyticinc.com.

MATLAB is a registered trademark of The MathWorks. Catalytic Inc. acknowledges trademarks or registered trademarks of other organizations for their respective products and services.