The OCP Open Domain-Specific Architecture (ODSA) Subproject Makes Significant Gains in Chiplet-based Architecture, Design and Industry Collaboration

Amsterdam, The Netherlands, Sept. 26, 2019 (GLOBE NEWSWIRE) -- The Open Compute Project (OCP) announced today significant progress by the Open Domain-Specific Architecture (ODSA) subproject in the development of a chiplet-based architecture.  Since its charter in March 2019 within OCP, the ODSA subproject has made critical steps in defining and developing a chiplet-based architecture with the introduction of new interfaces, link layers, a marketplace and exchange, and an early proof-of-concept.

Decades of progress with general-purpose CPUs have slowed, while performance requirements of workloads have catapulted, driving significant demand in domain-specific accelerators. Chiplet-based designs that combine multiple die into a single package can reduce the development time and manufacturing costs for accelerators. According to a preliminary research report by IHS/Informa, the aggregate market for chiplets is projected to be almost $3B by 2024 and grow to $10B in 2030..

The ODSA subproject’s mission is to define an open interface and architecture that enables the mixing and matching of silicon chiplets from different vendors via an open marketplace onto a single SoC. To achieve this goal, multiple working groups within the ODSA have been established: 

  • The ODSA PHY interface group is tasked with defining a simple, open, flexible data-rate interface between chiplets. This group has produced an objective analysis of multiple inter-chiplet PHY technologies published at a recent peer-reviewed conference. The group has also defined a new Bunch of Wires (BoW) interface. The interface aims to combine ease of design with process portability and suitability for use in low-cost packaging technologies. The group has released a 0.7 BoW specification today at the OCP Regional Summit. Group members include participants from Aquantia, Avera Semi, Netronome and zGlue.
  • The ODSA Proof of Concept (PoC) group is tasked with validating the technology proposals from the program. Equally important, the PoC group is working to understand and test business and technical issues around having multiple companies collaborating and sharing responsibility on a chiplet solution. As a first step, the group is working to deliver a software development prototype by the end of the year involving multiple small interoperable, interchangeable boards, which would model a single-chip solution based on chiplets. Group members include participants from Achronix, Cisco, Facebook, Netronome, NXP and zGlue.
  • The ODSA Business Working Group is tasked with defining a workflow and business processes to enable companies to assemble products from the marketplace. As a first step, the group has released a 0.9 specification for a Chiplet Design Exchange (CDX). The CDX specification is an extension of ZEF format developed by zGlue to describe chiplets and defines data sharing. Group members include participation from zGlue Inc, Ayar Labs, Microsoft Azure, and Netronome.

“We are observing new architectures emerging that solve for rapidly changing workloads currently not well-fit for a traditional large-scale integration approach to design. These new architectures provide improved approaches to rapidly and cost-effectively develop workload-specific products. We believe an open chiplet-based architecture being developed within the ODSA community provides a pathway towards achieving these goals, while enabling continued performance gains,” said Aaron Sullivan, director of hardware engineering at Facebook.

Since OCP launched the ODSA subproject, they have hosted three workshops, in conjunction with Samsung, Intel and IBM, to bring together a comprehensive ecosystem of chip designers, software and hardware architects, and cloud/data center operators to collaborate and solve a wide range of issues, from PHY specifications and protocols, link layers, chip and board design, to new business and licensing considerations.

“Since 2015, we’ve been successfully partnering with the OCP community to reimagine cloud hardware making it more efficient, flexible and scalable,” Kushagra Vaid, Distinguished Engineer, Azure Hardware Infrastructure, Microsoft. “Creating new open standards for chiplet-based architecture and interoperability is an important step forward to enabling emerging applications, like Machine Learning, that require compute resources at unprecedented scale.”

“OCP has become the community of choice for collaboration. The open architecture for domain-specific accelerators being developed by the ODSA subproject can bring the benefits of disaggregation to the world of SoCs. In just a short time, the ODSA has come with specifications, a chiplet marketplace and exchange, and an early prototype. We are very thankful for the tremendous support from participating companies for the excellent work being performed by the ODSA,” said Bill Carter, CTO of OCP.

Learn more about the ODSA at the OCP Regional Summit in Amsterdam, the Netherlands during the ODSA Session on September 26, and at the ODSA mini-workshop on September 27:

For additional information and how to get involved, please visit the project portal and join the mailing list.

About OCP

The Open Compute Project Foundation was initiated by Facebook in 2011 with a mission to apply the benefits of open source and open collaboration to hardware and rapidly increase the pace of innovation in, near and around the data center’s networking equipment, general purpose and GPU servers, storage devices and appliances, and scalable rack designs. OCP’s collaboration model is being applied beyond the data center, helping to advance the telecom industry & EDGE infrastructure. 

 

Dirk Van Slyke
Open Compute Project Foundation
303-999-7398
dirkv@opencompute.org
Featured Video
Jobs
Sr. Silicon Design Engineer for AMD at Santa Clara, California
CAD Engineer for Nvidia at Santa Clara, California
Design Verification Engineer for Blockwork IT at Milpitas, California
Senior Platform Software Engineer, AI Server - GPU for Nvidia at Santa Clara, California
Senior Firmware Architect - Server Manageability for Nvidia at Santa Clara, California
Upcoming Events
SEMICON Japan 2024 at Tokyo Big Sight Tokyo Japan - Dec 11 - 13, 2024
PDF Solutions AI Executive Conference at St. Regis Hotel San Francisco - Dec 12, 2024
DVCon U.S. 2025 at United States - Feb 24 - 27, 2025



© 2024 Internet Business Systems, Inc.
670 Aberdeen Way, Milpitas, CA 95035
+1 (408) 882-6554 — Contact Us, or visit our other sites:
AECCafe - Architectural Design and Engineering TechJobsCafe - Technical Jobs and Resumes GISCafe - Geographical Information Services  MCADCafe - Mechanical Design and Engineering ShareCG - Share Computer Graphic (CG) Animation, 3D Art and 3D Models
  Privacy PolicyAdvertise