Resources

BlogOptical Transceivers

Optical Interconnect Technology Analysis: LPO, NPO, CPO

Optical Interconnect Technology Analysis: LPO, NPO, CPO

Exploring optical interconnects for AI data centers: LPO for low-power, short-distance links, NPO for high-density, near-package connections, and CPO for ultra-high-bandwidth co-package solutions. These three optical interconnect technologies work together to improve bandwidth, power consumption, and system efficiency.

BlogAI Networking

How to Deploy NVIDIA DGX Spark

How to Deploy NVIDIA DGX Spark

How to Connect and Deploy NVIDIA DGX Spark? This article details direct dual-machine connections, multi-machine interconnects, and their applicable scenarios to help build efficient AI computing clusters.

BlogInfiniBand Networking

NVIDIA BlueField-4 DPU Powers Gigascale AI Factories

NVIDIA BlueField-4 DPU Powers Gigascale AI Factories

NVIDIA has released the BlueField-4 DPU, which integrates a 64-core Grace CPU and an 800Gb/s ConnectX-9 SuperNIC. Through the DOCA framework, it achieves integrated acceleration of computing, networking and security, providing system-level performance and security upgrades for hyperscale AI factories and data centers.

BlogAI Networking

In-Depth Analysis of Cisco N9100 AI Data Center Switch

In-Depth Analysis of Cisco N9100 AI Data Center Switch

Cisco N9100 AI Data Center Switch, based on NVIDIA Spectrum-X, provides a high-performance, open, flexible, and unified O&M AI networking solution suitable for new clouds, sovereign clouds, and large-scale AI clusters.

BlogAI Networking

The Performance of NVIDIA DGX Spark

The Performance of NVIDIA DGX Spark

NADDOD conducted performance experiments on the NVIDIA DGX Spark which verifies its powerful AI computing capabilities. It provides 1 PFLOP and supports local inference and fine-tuning of models with up to 200 billion parameters. This will completely revolutionize developers' AI workflows.

BlogAI Networking

In-depth Analysis of the Differences Between NVIDIA ConnectX-8 SuperNIC and ConnectX-7

In-depth Analysis of the Differences Between NVIDIA ConnectX-8 SuperNIC and ConnectX-7

In-depth analysis of the core differences between the NVIDIA ConnectX-8 SuperNIC and ConnectX-7, from 800G network performance and PCIe Gen6 architecture to DDP data placement and SHARPv4 aggregated computing, comprehensively revealing how ConnectX-8 is reshaping the interconnect architecture of AI data centers.

BlogAI Networking

NVIDIA Vera Rubin NVL144 : Next-Generation High-Performance Computing Platform

NVIDIA Vera Rubin NVL144 : Next-Generation High-Performance Computing Platform

NVIDIA Vera Rubin NVL144 showcases the architecture and scale of a new generation of high-performance computing platforms. Through the collaborative design of Rubin GPUs and Vera CPUs, it achieves outstanding computing power and system efficiency, providing a solid foundation for future large-scale computing.

BlogInfiniBand Networking

NVIDIA NVQLink Introduction：Connecting Quantum and GPU Computing

NVIDIA NVQLink Introduction：Connecting Quantum and GPU Computing

In-depth analysis of NVIDIA NVQLink technology, understanding how to extend GPU computing power to quantum processors through high-bandwidth, low-latency interconnects to build accelerated quantum supercomputers.

BlogAI Networking

NVIDIA RTX PRO Server Powered by ConnectX-8 SuperNIC Builds Next-Generation AI Architecture

NVIDIA RTX PRO Server Powered by ConnectX-8 SuperNIC Builds Next-Generation AI Architecture

NVIDIA RTX PRO Server is equipped with ConnectX-8 SuperNIC, providing 800Gb/s high-speed interconnection and Blackwell GPU computing power to meet high-performance computing needs such as AI reasoning, rendering, and digital twins.