MathWorks move model-based design into SoCs with Faraday

MATLAB and Simulink Speed Simulations by 200x and Reduce Gate Count by Half
By Nick Flaherty www.flaherty.co.uk


MathWorks is succcessfully moving its MATLAB and Simulink model-based deign tools more into silicon design. Faraday Technology is using Model-Based Design to accelerate the development of SoCs, including development of its NAND Flash Controller error correcting code (ECC) engine. MATLAB and Simulink have helped Faraday accelerate development of more efficient designs by completing system-level simulations 200 times faster, increasing throughput performance by 15%, and cutting gate count by up to 57%.

Stateflow chart of the error correcting code (ECC) engine for Faraday’s NAND flash controller.

Faraday is using this to capitalise on the move to advanced SoC designs. To meet this demand, Faraday established a faster and more cost-efficient workflow to develop memory controllers for microprocessor subsystems. Faraday adopted Model-Based Design to accelerate SIP development, explore system-level design alternatives, and improve communication among engineers. This design approach has helped reduce gate count, and shorten the development cycle for memory controllers and other modules, and, as a result, differentiate itself through lower cost to its customers.
“Model-Based Design provided an efficient and cost-effective way to improve silicon intellectual property development by enabling us to rapidly identify the best design configurations and get products to market faster,” said Ken Chen, ESL Methodology Manager at Faraday. “The Simulink environment is ideal for integrating, simulating, and exploring design architectures. The simulations are up to 200 times faster than RTL simulations, and Simulink models can be easily converted to HDL code for FPGA prototyping.”

Optimising the gate count of the ECC block

Faraday engineers used MATLAB, Simulink, and Stateflow to model and simulate their system-level designs, and used Simulink Coder and Simulink HDL Coder to automatically generate code from their models. This workflow enabled Faraday to shorten their design process as they moved from architecture design to hardware and software implementation. Faraday has used Model-Based Design for DDR and flash controller projects and delivered the SIP designs on schedule. The engineering team is now focused on further accelerating development by reusing and adapting their existing models on new projects.
“As integrated circuit manufacturers continue to rely on SIP providers for SoC and ASIC components, SIP engineers need the ability to rapidly deliver designs that meet their customers’ requirements,” said Ken Karnofsky, senior strategist for signal processing applications, MathWorks. “Faraday’s use of MATLAB and Simulink showcases the effective use of Model-Based Design to accelerate development, reduce costs, and simplify the integration of SIP modules into SoC designs.”