At a glance
- Company
- Huawei
- Announcement
- Product
- Location
- Shanghai, China
Key figures
- Processor scale supported by architecture
- million scale
- Cards in Atlas 950 SuperCluster being deployed
- 256,000 cards
People
- Eric XuHuawei's Rotating Chairman
Products and programs
The Peerium Computing Architecture achieves strong scaling to the million-processor level through nested parallelism, unified memory addressing, and peer interconnect. It breaks through the Turing paradigm with the introduction of Nested BSP (Nested Bulk Synchronous Parallel), extends the von Neumann single-machine architecture, and overturns the master–slave architecture that has prevailed for decades, so that a million processors truly become one larger computer.
UnifiedBus (UB) is the key interconnect technology that makes the Peerium Computing Architecture possible. Built on a single open protocol, UB is a high-speed bus that scales without limit to connect CPUs, NPUs, memory, SSDs, network interface cards (NICs), and switches. With UB, peer interconnect is achieved across compute, storage, and networking.
The Atlas 950 SuperPoD and SuperPoD-based SuperClusters are the first-generation product built on the Peerium Computing Architecture. An Atlas 950 SuperCluster with 256,000 cards is already being deployed, and the Atlas 960 system based on near-packaged optics (NPO) is currently under testing.
Eric Xu, Huawei's Rotating Chairman, said, "In the AI era, Huawei is drawing on the Peerium Computing Architecture we pioneered to continuously build the SuperPoDs and SuperPod-based SuperClusters that meet customer needs for training and inference, making computing power available everywhere and intelligence accessible to all."
Hashtag: #Huawei
The issuer is solely responsible for the content of this announcement.
Questions around this news
What is the Peerium Computing Architecture?
Peerium is a computing architecture announced for AI workloads. The release says it is designed to let processors at the million scale operate as one computer, using nested parallelism, unified memory addressing, and peer interconnect. It also introduces Nested BSP and extends the von Neumann single-machine architecture.
What role does UnifiedBus play in the architecture?
UnifiedBus is the interconnect technology that enables the architecture’s peer connections. The release describes it as a high-speed bus based on one open protocol, designed to link CPUs, NPUs, memory, SSDs, network interface cards, and switches across computing, storage, and networking.
Which products use the Peerium Computing Architecture?
The release identifies the Atlas 950 SuperPoD and SuperPoD-based SuperClusters as the first-generation products built on Peerium. It says an Atlas 950 SuperCluster with 256,000 cards is being deployed, while the Atlas 960 system, which uses near-packaged optics, is under testing.
What is Nested BSP?
Nested Bulk Synchronous Parallel is a parallel-computing model in which work is organized into computation phases separated by synchronization points. The “nested” approach allows parallel tasks and synchronization phases to be arranged within larger ones, providing a way to structure coordination across levels of a system.
Earlier from Huawei
- Huawei Unveils the Industry's First NPO Optical Engine with Built-in Light Source, Hi-ONE, Leading the Way in High-Speed Optical Interconnect Technology Innovation
- HUAWEI Mate XT 2 | ULTIMATE DESIGN Debuts as the First Smartphone to Feature LogicFolding Tau (τ) Chip and Reach Mass Commercialization
- Huawei He Tingbo Named to TIME'S Fourth Annual List of The TIME100 AI
- Now is Your Moment: Huawei Unlocks New Possibilities for Every Moment Through All-Scenario Innovation
- Huawei Announces Wi-Fi 7 Patent Licensing Rates, Reaffirms Commitment to Transparent and Reasonable Licensing
