GigE Vision

RoCEv2 Integration into GigE Vision Standard: A Leap Forward in Imaging Technology


Introduction:

GigE Vision is an A3 Vision & Imaging standard that was introduced in 2006 with the aim of standardizing communication between applications and machine vision devices over conventional Ethernet technologies. The machine vision industry quickly recognized the significant advantages offered by GigE Vision, including enhanced flexibility, cost-effectiveness, and improved performance. Over time, the specification has evolved to cater to the expanding requirements of machine vision systems. In the last major update in 2011 to version 2.0, the specification was enhanced to accommodate even greater functionality. Today, GigE Vision stands out as the most versatile transport layer for machine vision applications.

The GigE Vision Technical Committee is excited to announce our ongoing work on the next major release of the standard version 3.0, which aims to offer significant benefits for both machine vision and industrial automation. This update builds upon the advancements in Ethernet hardware, particularly the network interface card (NIC), to enhance the standard's capabilities. Our primary focus in this update is the integration of RoCEv2 (RDMA over Converged Ethernet version 2) for data streaming, complementing existing streaming protocol and bringing about notable improvements in performance, reliability, and efficiency. It's important to note that RoCE is being introduced as an addition rather than a replacement, emphasizing our commitment to enhancing the standard's capabilities without disregarding existing functionalities. Furthermore, it's significant to highlight that leading NIC vendors have integrated RoCEv2 offloading into their premium products, underscoring the technical prowess and strategic adoption of this networking protocol.

This new major release will mark a pivotal moment in the dynamic world of machine vision, promising to redefine the data transmission process. RoCEv2 brings unparalleled speed, minimal latency, and an overall boost to system performance. As we embrace this advancement, we anticipate a transformative impact on the capabilities and possibilities within the realm of imaging technology.

Understanding GigE Vision:

GigE Vision, an industrial camera interface standard, has been widely adopted in various applications, including manufacturing, robotics, medical imaging, and more. It utilizes Gigabit Ethernet technology to transmit high-quality images and video over long distances with low latency. While GigE Vision has been successful, the advent of RoCEv2 brings a new layer of capability to this standard.

Introduction to RoCEv2:

RoCE, or RDMA over Converged Ethernet, is a protocol that enables direct memory access from a device to a computer without involving the operating system. This technology was initially developed to accelerate data transfer in high-performance computing environments. RoCEv2, the second version of this protocol, refines and enhances its predecessor, providing improved efficiency and flexibility. Notably, RoCEv2 operates atop the UDP/IP protocol stack, ensuring seamless integration and compatibility within Ethernet networks while enhancing overall functionality.

Key Benefits of RoCEv2 in GigE Vision:

  • Low CPU Utilization:
    The efficiency of RoCEv2-capable NICs is notable in their ability to offload data transfer tasks from the CPU. This reduction in CPU involvement in the data transfer process ensures that system resources are freed up, allowing them to be utilized for other tasks such as image processing. Consequently, this leads to enhanced overall performance.
  • Scalable Throughput:
    RoCEv2 NICs offer a range of physical links, spanning from 10Gbps to 400Gbps. This broad spectrum of speed options significantly amplifies GigE Vision adaptability to varied network demands, ensuring scalability across a diverse array of applications.
  • Enhanced Reliability in High Throughput:
    RoCEv2 enhanced the reliability of GigE Vision through its robust error handling capabilities. By delegating error detection and recovery tasks to dedicated hardware, the protocol ensures a smoother and more consistent data transfer. This proactive approach minimizes disruptions in critical applications, enhancing stability and reducing the risk of interruptions.
  • Low Latency:
    RoCEv2 introduces low latency, a critical factor in machine vision applications where real-time data processing is essential. This reduction in latency enhances the responsiveness of the imaging system, making it ideal for applications such as robotics and automation where split-second decisions are crucial.
  • Compatibility and Interoperability:
    An essential prerequisite for the development of the upcoming major release is to preserve compatibility with existing devices. The current GigE Vision control protocol will be retained as the mechanism for device discovery and configuration. This strategic choice ensures a seamless transition for industries and applications currently reliant on GigE Vision, minimizing the need to significantly overhaul existing software while providing users with the flexibility to leverage new streaming capabilities.

Conclusion:

The addition of RoCEv2 to the GigE Vision standard represents a leap forward in imaging technology. With its focus on low CPU utilization, high throughput, low latency, and compatibility, RoCEv2 addresses the evolving demands of machine vision and industrial automation. As industries continue to push the boundaries of what is possible with imaging systems, the integration of RoCEv2 into GigE Vision sets the stage for a new era of efficiency, speed, and reliability in the world of industrial imaging.

If you want be part of shaping the future of industry standards and advancing machine vision and industrial automation technologies, we invite you to join the GigE Vision Technical Committee and contribute to the development of this new major release.

Eric Bourbonnais
Chair of the GigE Vision Technical Committee