NetSpeed Systems Inc., the supplier of on-chip network IP that is streamlining SoC design, announced the release of its Gemini 2.0 product, the industry’s first fully configurable cache-coherent network-on-chip IP. The new release adds a number of advanced features that make it faster and easier for SoC architects to configure interconnect solutions that are highly optimized for specific applications.
“A major challenge for next generation SoC designs is the explosion of compute clusters and demand for ultra-low latency memory access,” said Sundari Mitra, CEO of NetSpeed Systems. “Gemini 2.0 fills the need for manufacturers that want to use innovative cache-coherent designs to squeeze more performance from their designs. Gemini offers them far more configurability than other coherency solutions with better performance, power, and area.”
“Configuring coherent systems has traditionally required specialized expertise that few SoC designers possess,” said Fred Weber, former CTO of AMD and member of NetSpeed’s board of directors. “However, NetSpeed has developed a sophisticated and resilient method of configuring cache in coherent systems that uses automation and algorithms to ensure that the resulting interconnect is high performance, functionally correct and deadlock-free.”
Gemini 2.0 is a development platform for architecting efficient cache-coherent interconnects for a wide variety of SoC applications. It uses a requirements-driven design approach that enables architects to work at a higher level of abstraction. Based on system requirements such as cache capacity and total coherent bandwidth, Gemini adds critical coherency components and customizes them before placing them in the interconnect. Gemini uses a distributed architecture and an innovative directory-based design that enables architects to scale both the interconnect fabric and the coherency components up or down independently. The latest version of Gemini also offers NetSpeed Pegasus - an optimized last-level cache, and an accelerator for I/O-coherent traffic — which is ideal for GPU-like IPs.
NetSpeed’s Gemini IP was first introduced in the fall of 2013 to offer customers a highly configurable, latency-optimized solution for today’s heterogeneous architecture designs. It is currently used by Altera and other early adopter customers in advanced process nodes.
Gemini is a scalable high-performance on-chip network IP solution that provides full cache coherency for small and large SoC designs. It supports all three levels of coherent traffic including cache-coherent, I/O-coherent & non-coherent traffic – in a single on-chip network. NetSpeed Gemini's unique architecture allows it to scale performance to match increasing coherency bandwidth. This allows NetSpeed Gemini to be used as NoC platform for entire product families. The underlying hardware elements, like the coherency controller, coherency directory, and NoC router are designed for supporting higher throughput – Terabits per second – with a lower footprint. Gemini delivers high performance and significant time-to-market advantages to SoC designers for a wide range of markets including mobile, networking and enterprise computing.
About NetSpeed Systems, Inc.
NetSpeed Systems provides scalable, coherent on-chip network IPs to SoC designers for a wide range of markets from mobile to high-performance computing and networking. NetSpeed's on-chip network platform delivers significant time-to-market advantages through a system-level approach, a high level of user-driven automation and state-of-the-art algorithms. NetSpeed Systems was founded in 2011 and is led by seasoned executives from the semiconductor and networking industries. The company is funded by top-tier investors from Silicon Valley. It is based in San Jose, California and has additional research and development facilities in Asia.
A major challenge for next generation SoC designs is the explosion of compute clusters and demand for ultra-low latency memory access. Gemini fills the need for manufacturers that want to use innovative cache-coherent designs to squeeze more performance.