The MathWorks has introduced Fixed-Point Toolbox 2, which the company said provides enhanced floating-to-fixed-point conversion capabilities and accelerated fixed-point MATLAB algorithms able to execute at compiled C-code speed.
The new version gives design engineers “a cohesive workflow for optimizing and verifying embedded algorithms entirely within MATLAB,” according to technical marketing manager Houman Zarrinkoub. Benefits include faster design iterations and elimination of translation errors.
Zarrinkoub noted that embedded signal processing and control systems often require fixed-point algorithms for implementation on digital signal processors, microcontrollers, application-specific integrated circuits, and field-programmable gate arrays. “A major challenge is maintaining the correct behavior of an algorithm when converting it from floating-point to fixed-point representation,” he said.
Fixed-Point Toolbox 2 provides tools for data logging and data-type override that are intended to streamline the conversion process and ensure consistent algorithm behavior in both representations. An accelerated simulation mode increases the execution speed of fixed-point MATLAB algorithms by factors of up to a thousand, according to Zarrinkoub.
“Fixed-point development usually involves managing a design expressed in several different languages, such as MATLAB, C, assembly, and HDL,” he said. “This is a time-consuming and error-prone process.” Fixed-Point Toolbox 2 provides tools for design validation and trade-off analysis on fixed-point algorithms in MATLAB. It also facilitates the use of fixed-point MATLAB algorithms within Simulink for system simulation, verification, and automatic generation of embeddable C code. “Because the algorithms have the same fixed-point representation across MATLAB and Simulink, design engineers can complete their designs in a single environment,” Zarrinkoub said.