The chips are great. Whether it’s the fried potato variety or the silicon wafer type in Central Processing Unit (CPU) or Graphics Processing Unit (GPU) variants. But processing always has a shelf life and we all know the wider effects of Moore’s law Nowadays, we’ve miniaturized the micron size of transistors to such an extent that we’ve had to start packing them together to get more power… and so on. But this is a brief history of chips as we knew them…oOur next future in this space will of course be defined by photonic integrated circuits which use light with the ability to process at high speed with low latency and (especially in the era of sustainability-focused computing) low power.
Photonics is at the heart of the Innovative Optical and Wireless Network (IOWN) initiative and we now see NTT Corporation (NTT) and Red Hat, Inc., in collaboration with Fujitsu as well as capitalization enthusiasts Nvidia – will the companies now joining forces with has jointly developed a solution to enhance and expand the potential of real-time artificial intelligence (AI) data analysis at the edge.
The Innovative Optical and Wireless Network (IOWN) is an initiative for future communications infrastructure aimed at creating a smarter world using cutting-edge technologies. IOWN is built around photonics technology for ultra-high capacity, ultra-low latency and ultra-low power consumption.
Today, most of our devices and technologies (phones, watches, games, sensors, PCs, servers, networks, etc.) use electronics to process and transmit information. IOWN will use optical technologies to transform these electronic connections into photonic connections, increasing transmission speeds and improving responsiveness while consuming extremely low energy levels.
IOWN involves devices, networks, and information processing infrastructure built on optical and other technologies to provide high-speed, high-capacity communications and computing resources. IOWN consists of three key technology areas:
- The All Photonics Network (APN), which applies optical technology: APNs seek to overcome the limitations of the existing network by converting all signals into optical signals, creating a network with higher capacity, lower latency and lower consumption. lower energy than today.
- Digital Twin Computing (DTC) for advanced real-time interaction between objects and people in cyberspace.
- The Cognitive Foundation (CF), which effectively deploys various ICT resources, including those above.
Using technologies developed by the IOWN Global Forum and built on the basis of OpenShift Red Hat For the power of hybrid cloud application platform powered by Kubernetes, this solution received a IOWN Global Forum Proof of Concept (PoC) recognition for its real-world viability and use cases.
As AI, sensing technology, and networking continue to accelerate, using AI analytics to evaluate and sort inputs at the network edge will be critical, especially more than data sources are growing almost daily. However, using AI analytics at scale can be slow and complex, and may be associated with higher maintenance costs and software maintenance to integrate new AI models and additional hardware . With edge computing capabilities emerging in more remote locations, AI analytics can be placed closer to sensors, reducing latency and increasing bandwidth.
This solution includes the IOWN All-Photonics Network (APN) and data pipeline acceleration technologies in IOWN’s Data Centric Infrastructure (DCI). NTT’s accelerated data pipeline for AI adopts Remote Direct Memory Access (RDMA) over APN to efficiently collect and process large amounts of edge sensor data. Red Hat OpenShift Container Orchestration Technology3 provides greater flexibility to manage workloads within the accelerated data pipeline in geographically distributed and remote data centers. NTT and Red Hat have successfully demonstrated that this solution can effectively reduce power consumption while maintaining lower latency for real-time AI analytics at the edge.
“The NTT Group, in close collaboration with its partners, is accelerating the development of IOWN to achieve a sustainable society. This IOWN PoC is an important step towards green computing for AI, which supports the collective intelligence of AI. We are further improving IOWN’s energy efficiency by applying photonic-electronic convergence technologies to an IT infrastructure. We aim to embody the sustainable future of net zero emissions with IOWN,” said Katsuhiko Kawazoe, senior executive vice president of NTT and president of the IOWN Global Forum, speaking at Mobile World Con2024 congress in Barcelona.
Proof of concept evaluated real-time AI analytics platform5 with Yokosuka City as the sensor installation base and Musashino City as the remote data center, both connected via APN. As a result, even when a large number of cameras were installed, the latency required to aggregate sensor data for AI analysis was reduced by 60% compared to conventional AI inference workloads . Additionally, IOWN PoC testing demonstrated that the power consumption required for AI analysis for each edge camera could be reduced by 40% compared to conventional technology.
This real-time AI analysis platform allows the GPU to be scaled to accommodate a larger number of cameras without the CPU becoming a bottleneck. According to a trial calculation, assuming that 1,000 cameras can be installed, energy consumption should be reduced by a further 60%. The highlights of the proof of concept of this solution are as follows.
“Over the past several years, we have worked through the IOWN Global Forum to lay the groundwork for open source-powered AI innovation and deliver technologies that help us make smarter choices for the future. This is important and exciting work, and these results help prove that we can create AI-based solutions that are sustainable and innovative for businesses around the world. With Red Hat OpenShift, we can help NTT deliver large-scale AI data analysis in real-time and without limitations,” said Chris Wright, CTO and SVP of Global Engineering at Red Hat and Board Director of the IOWN Global Forum.
Features here include:
- Accelerated data pipeline for AI inference, provided by NTT, uses RDMA over APN to directly fetch large-scale sensor data from local sites to the memory of an accelerator in a remote data center, thereby reducing protocol management overhead in the conventional network. It then completes the processing of AI inference data within the accelerator with less CPU control overhead, thereby improving the energy efficiency of AI inference.
- Real-time, large-scale AI data analysispowered by Red Hat OpenShift, can help Kubernetes operators minimize the complexity of implementing hardware accelerators (GPU, DPU, etc.), enabling improved flexibility and easier deployment across disaggregated sites, including remote data centers.
- This PoC uses NVIDIA A100 Tensor Core GPUs and NVIDIA ConnectX-6 network cards for AI inference.
The collective hope here is that technology companies contribute to achieving a sustainable and smarter society by applying technologies for smarter disaggregated computing.
