Cooperate with the AI Hardware Center of IBM Research Institute to solve the challenges of using innovative design methods to develop new AI chip architectures
Mountain View, California December 28, 2020--
Key points:
Synopsys, as the chief EDA and IP partner of the IBM AI Hardware Research Center, has achieved significant improvements in silicon verification and performance through its unique cooperation with IBM
Synopsys provides design, verification and IP solutions and technical expertise focused on the development of new AI-specific hardware architectures
Cross-industry cooperation to deal with key technical challenges to achieve the performance expansion and energy efficiency improvement required for the expansion of AI in various application scenarios and use cases
Synopsys (Synopsys, Inc., NASDAQ: SNPS) recently announced that its continued cooperation with the AI Hardware Center of IBM Research has entered a new stage to jointly advance the critical chip architecture and design methods in the next generation of AI chips Development. Synopsys works closely with IBM to implement the latest AI hardware center technology in a full-chip solution and realize its commercialization in the near future. The two parties have carried out a unique cooperation since last year, making full use of the rich expertise of IBM's famous research institutions, and combining the support of multiple business partners and academic and government institutions.
The overall goal of Synopsys' cooperation with IBM is to continue to double the AI computing performance every year in the next ten years or even longer. In order to achieve this goal, the two companies are working hard to redesign hardware around AI in order to expand the scope of AI use, thereby solving many problems faced by enterprises and the entire world. The cooperation between the two parties includes the development of new computing accelerators, technologies and architectures specifically designed and optimized for AI computing.
Mukesh Khare, vice president of hybrid cloud at IBM Research Institute, said: “AI and hybrid cloud will play an important role in the next generation of enterprise computing and extended AI, and the new hardware solutions for it are IBM Research Institute’s future development plans for AI. In order to achieve this goal, we need to build a new type of AI hardware accelerator to increase computing power without increasing energy consumption. In addition, the development of a new AI chip architecture will allow companies to develop a hybrid cloud Dynamically run large AI workloads in China. In this work, Synopsys' unparalleled rich experience and technical level will provide us with a great help."
The AI Hardware Center has implemented multiple tapeouts and test chips designed for advanced process manufacturing nodes to support its active roadmap, including increasing the AI computing performance by 1,000 times by 2029, which also means that the AI processor core Performance will increase 2.5 times every year, and IBM Research has achieved a double increase in the first year.
Synopsys is deeply involved in the project, including technical cooperation and cooperation between engineers and IBM researchers, and will focus on solving major challenges in the design, verification and manufacturing of complex AI chips. Specifically, Synopsys will bring expertise in three main areas:
Use Synopsys 3DIC Compiler, Fusion Design Platform and VerificationContinuum platform to implement multi-die chip integration in packaging, silicon design and verification, including the use of the latest functional verification, prototyping and hardware acceleration systems to solve the size of the design in development And scale issues, as well as support for hardware and software co-design and co-analysis methods.
In terms of silicon engineering, software is provided to solve major manufacturing and productivity challenges brought about by leading process technologies (such as the use of novel materials, all-round 3D gate stack architecture, the source of EUV technology and mask creation). Our design process collaborative optimization (DTCO) solutions and excellent technical support can provide more technical options and help achieve the global best.
In terms of silicon IP, it can meet the processing, memory performance and real-time connection requirements of AI chips, and provide a wide range of silicon-proven DesignWareIP products, such as LPDDR5 and PCI Express5.0, to meet various application requirements.
Arun Venkatachar, Vice President of Artificial Intelligence and Central Engineering at Synopsys, said: “The cooperation with IBM provides Synopsys with a unique opportunity to participate in connecting the entire semiconductor value chain. To realize the grand plan of the IBM Research Institute, a new approach is needed. Designing AI hardware requires innovative strategies from tools and IP to workflow and manufacturing. Cooperation with the AI Hardware Center of IBM Research Institute will provide us with an important platform to build the future of AI chip design with our partners."
About Synopsys
Synopsys (Synopsys, Inc., NASDAQ: SNPS) is a Silicon to Software ("chip to software") partner of many innovative companies that are committed to developing the electronic products and software application. As the world's 15th largest software company, Synopsys has long been a global leader in the fields of electronic design automation (EDA) and semiconductor IP, and it has also played an increasingly important role in software security and quality solutions. Whether you are a system-on-chip (SoC) designer who creates advanced semiconductors, or a software developer who writes applications that require the highest security and quality, Synopsys can provide the solutions you need and help you launch innovative , High-quality and safe products.