Intelligent clock network platform for SoC designs
The Maestro Intelligent Clock Network platform is designed to provide critical clocking capabilities that address complex clock distribution requirements in modern SoCs, including AI applications in the cloud and at the edge. The Maestro clocking solutions, says the company, dramatically improve the entire chip’s power, performance, and area in a wide variety of markets.
Influencing every aspect of the system from timing closure to peak power demand, and from chip architecture to layout, the clock is one of the most complex signals in an SoC. The Maestro platform is designed to address the SoC-level clock distribution challenges experienced with designs of all sizes and technology nodes.
“The complexity of today’s SoC designs, especially with the emergence of new AI chip architectures, is mind boggling,” says Mo Faisal, President and CEO of Movellus. “We see large scale designs with thousands of processors on a chip. We see IoT designs with nanowatt power requirements. System designers want to do things today that, quite frankly, would not be possible using traditional clock distribution solutions. Customers are very excited by the new capabilities intelligent clock networks bring. With the diversity of new and emerging architectures, there is a genuine and growing market need for them.”
Currently, says the company, clocking challenges are addressed by overdesigning the system, creating further costs including up to 30% to 50% overhead of the SoC power and area. Furthermore, these designs end up limiting maximum clock frequency and consequently limit performance.
The Maestro platform is designed to automate the development of robust clock network solutions with powerful new capabilities. It combines a clock architecture, software automation, and application-optimized IP to solve common clock distribution challenges.
Maestro is offered as the only platform that provides end-to-end all-digital clocking network solutions, which offer better power, performance and area (PPA) and superior operating characteristics compared to traditional solutions. Additionally, they can scale to high-performance multi-gigahertz frequencies while occupying only a small footprint.
Maestro incorporates the company’s TrueDigital construction including Clock Generator Modules (CGM), Smart Clock Modules (SCM), and Phase Shift Modules (PSM). CGMs are used for a variety of functions including clock generation and improved signal quality along the clock path. SCMs automatically sense and compensate for skew in the SoC clock path during runtime. PSMs reduce peak power and reduce voltage noise thereby relaxing voltage margins.
Maestro also has the ability to create a virtual mesh, which, says the company, can be thought of as a “network-on-chip (NoC) for the clock.” The virtual mesh can sense and compensate for on-chip variation and inter-block interactions globally. It enables automatic self-regulation of clocking throughout the SoC. Maestro enables highly efficient ultra-low power operation and supports high frequency multi-GHz performance. It eliminates the need to overdesign the system to compensate for on-chip variation or skew.
Three reference configurations of the Maestro Intelligent Clock Network platform are available to address specific requirements based on the complexity of the application: Maestro AI, Maestro Cloud, and Maestro Phases. All use TrueDigital construction to generate RTL code that precisely meets application specifications.
Their fully synthesizable architectures enable rapid RTL-to-GDS implementation and optimization. Maestro is designed to work seamlessly with existing digital tools and methodologies to reduce design complexity and significantly improve development time.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
