OSN9800 TNU1UCXCS Hyper Converged Programmable Board for Next-generation Intelligent Network Architecture

Yantai Boxin International Trading Co., Ltd. is a professional comprehensive trading service provider for communication equipment.With warehouses established in Yantai and Hong Kong, we have built a complete and localized network for delivering communication products and services.In 2018, Yantai Boxin International Trading Co., Ltd. set up its international sales department in Yantai, China.Based in China, we operate in emerging markets across Southeast Asia, India, the Middle East, Africa, Europe, and the United States. 

Yantai Boxin International Trading Co., Ltd. supplies products to various telecom operators, meeting their needs for supplementing and expanding 2G/3G/4G/5G network stations.Our current equipment portfolio covers brands such as Nokia, Ericsson, Alcatel, Huawei, ZTE, ZTT, and FiberHome. 

We expand our international market share by offering high-quality products, excellent service, competitive pricing, and timely delivery.Feel free to contact us for more information.

Description du produit

1.Product Overview

TNU1UCXCS is a strategic research prototype system on the OSN 9800 series optical transmission platform, aimed at exploring the forefront of"Post by Neumann architecture" and"network computing storage" integration.This product goes beyond the scope of traditional wavelength division boards or programmable networks, and is a physical experimental carrier for underlying fusion innovation in fields such as computing architecture, photon integration, storage computing integration, and heterogeneous computing.Its core mission is not to solve current network problems, but to serve as an open infrastructure for exploring the fundamental paradigm of future information processing, aiming to provide revolutionary architectural prototypes and technological reserves for the form of information infrastructure after 2030.

2.Core positioning and basic research mission

Explore information processing architectures that go beyond existing paradigms

Aiming to challenge the classic von Neumann architecture of separating computing units, storage units, and communication units, this study explores the possibility of deep collaboration between optical transmission, near data computing, in memory processing, and other functions at the physical and architectural layers.

It is a comprehensive hardware experimental platform for verifying disruptive concepts such as"optical computing network" and"storage computing network integration".

Heterogeneous Fusion and Reconfigurable Computing Optoelectronic Base

The single board system integrates general-purpose computing cores (CPU/GPU class), programmable network processors (NP), programmable logic units (FPGA), and high-speed optical interconnect engines.

Support the dynamic reconstruction of the connection and collaboration modes of these heterogeneous computing and communication resources based on research objectives, in order to explore the optimal energy efficiency ratio and performance models.

Strategic reserve for long-term development of information science

To serve the long-term strategic layout in areas such as basic software, new computing, and advanced networks, with the goal of forming an independent knowledge system based on fundamental technologies.

The original innovation generated through this platform aims to reserve a batch of"seed technologies" that may define the next generation of game rules in future information industry competition.

Open and collaborative global top tier basic research hub

As a major infrastructure for basic research, it is open to the world's top theoretical computer scientists, architecture experts, and optophysicists to jointly challenge the fundamental problems of information science.

3.Core Technology Architecture and Experimental Characteristics

1.Physics and System Architecture

Model: TNU1UCXCS

Platform attribute: Frontier research and prototype verification platform for information architecture

Design philosophy:"Full stack openness, deep measurability, and cross layer collaboration".

2.Core heterogeneous fusion unit

Computing subsystem: including general-purpose multi-core processors, dedicated AI acceleration units, and programmable logic arrays, supporting dynamic offloading and mixed execution of computing tasks.

Storage subsystem: Integrated high bandwidth memory (HBM), new non-volatile memory, and customizable memory controller, supporting in memory computation, in memory search, and other near data operation experiments.

Communication and Interconnection Subsystem: Includes programmable network processors, high-speed SerDes, and 800G/1.2T coherent optical engines to achieve fully programmable interconnection within the board, between boards, and at the optical layer.

Collaborative management unit: provides a unified framework for resource abstraction, task scheduling, and performance monitoring.

3.Extreme Open Experiment Interface

Microarchitecture level access: Open the underlying interfaces such as microcode of key computing cores, timing parameters of memory controllers, and driving waveforms of optical modulators to researchers.

Cross layer performance probe: provides full stack, high-precision performance and energy consumption monitoring points from transistor switch activity, cache hit rate, to optical signal-to-noise ratio.

Pluggable experimental module: supports the insertion of third-party developed dedicated computing chips, new storage media, or optical modulation devices through standard interfaces.

4.Research supporting ecosystems

Full system cycle precise simulator: a software simulation environment that is completely consistent with the behavior of the physical platform, supporting exploration of ultra large scale architectures.

Automated benchmarking and validation framework: integrating a comprehensive testing suite and performance analysis toolchain covering computing, storage, and networking.

Open source research community and dataset: Establish a global open source research community around the platform, sharing architecture design, experimental data, and benchmark models.

4.Typical frontier research directions

Exploration of New Non Von Neumann Computing Architecture: Prototype Implementation and Evaluation of Data Flow Architecture, Neuromorphic Computing Architecture, and Quantum Classical Hybrid Computing Architecture in Optical Interconnection Scenarios.

Research on the integration of photon computing and electronic computing: Exploring the feasibility and system architecture of using optical computing units (such as optical matrix multipliers) as co processors to collaborate with electronic computing cores for specific tasks (such as AI inference and linear algebra).

Research on Data Centric Network Architecture: Exploring a New Paradigm of Directly Embedding Network Functions (such as Routing, Security Policies) into Intelligent Memory or Near Memory Processing Units to Achieve Data, Computing, and Processing Wherever Data is Found.

Research on the Fundamental Energy Efficiency Limits of Information Processing Systems: Measuring and analyzing the potential improvement space of new devices and architectures on system level energy efficiency at the complete system level across computing, storage, and communication, approaching theoretical limits.

Research on basic software: providing a low-level hardware experimental bed for researching new resource abstraction models, scheduling algorithms, and programming models for operating systems, compilers, and other system software.

5.Summary of Core Values

Defining the foundation of future information science: providing the most fundamental hardware experimental capabilities for achieving strategic transformation from"tracking imitation" to"source leadership" in the field of information.

Nurturing disruptive technology clusters: It is expected to incubate multiple original innovative technologies on a single platform that may disrupt the existing computing, storage, and networking industry landscape.

Cultivate the core competencies of strategic scientists: provide a practical environment for top researchers to master ultra complex, interdisciplinary heterogeneous systems, and cultivate"chief engineer" talents who define and design future complex systems.

Building long-term foundational advantages: By continuously investing and accumulating on such fundamental platforms, we aim to establish deep, broad, and difficult to replicate foundational technological advantages for the information industry.

6.Access and cooperation mechanism

Platform code: TNU1UCXCS.

Operating mechanism: Adopting the"Major Basic Research Projects under the Guidance of the Strategic Scientists Committee" model.Select research topics with transformative potential by an interdisciplinary committee of strategic scientists.

User qualification: Only applicable to core basic research teams from laboratories and top research universities, and research topics must have clear long-term, fundamental, and disruptive potential.

Achievement management: Research achievements (papers, open source code, prototype design) emphasize global academic influence and open source sharing, aiming to promote the progress of basic theories of information science for all mankind. The core intellectual property generated follows the management mechanism of major basic research achievements.