The article reviews NVIDIA's GPU architecture progression—from Volta's pioneering Tensor Cores through Turing, Ampere, Hopper, and the latest Blackwell and Rubin designs—highlighting key innovations, performance gains for deep learning, and related resource updates for AI practitioners.
At HPCA 2025, two standout papers—UniNDP, a unified compilation and simulation tool for near‑DRAM processing architectures, and WarpDrive, a GPU‑based fully homomorphic encryption accelerator leveraging Tensor and CUDA cores—demonstrate significant performance gains for AI workloads and privacy‑preserving computation.
The article explains how RDMA technologies—including InfiniBand, RoCE, and iWARP—provide high‑throughput, low‑latency, CPU‑free data transfer for massive generative AI model training, compares their architectures, and discusses modern network designs and load‑balancing strategies to optimize AI‑focused data‑center networks.
The article analyzes the networking challenges of AI super‑large models, comparing InfiniBand and RoCE technologies, and presents design guidelines for ultra‑scale, high‑bandwidth, low‑latency, and highly stable cluster interconnects to maximize GPU utilization and overall training efficiency.
This article explains how Remote Direct Memory Access (RDMA) technologies such as InfiniBand and RoCE bypass OS kernels to achieve ultra‑low latency and high bandwidth, discusses their hardware implementations, cost considerations, and their critical impact on large‑scale AI model training and HPC network design.
The article explains the Dragonfly network topology, its hierarchical structure, key parameters, routing algorithms (Minimal, Non‑Minimal, UGAL variants) and deadlock avoidance techniques, highlighting how modern data‑center networks address latency bottlenecks in high‑performance computing environments.
The 2021 Data Center High‑Quality Development Conference in Beijing showcased the release of industry indexes and low‑carbon white papers, awarded Alibaba Cloud for liquid‑cooled cloud computing innovations, and detailed China's rapid data‑center growth, emerging liquid‑cooling technologies, and collaborative standard‑setting efforts.
The article provides a comprehensive overview of GPU technology, its architecture, rapid market growth, segmentation by type, device and industry, cloud deployment trends, competitive landscape, and diverse applications ranging from high‑performance computing and AI to automotive, AR/VR, and IoT.
The article explains Nvidia's $6.9 billion acquisition of Mellanox, outlines Mellanox's history and product portfolio, and provides a detailed overview of InfiniBand architecture, network topologies, protocols, and related software stacks such as OFED, highlighting their importance for data‑center, HPC, and cloud environments.
Intel has reportedly offered $5.5 billion to acquire Mellanox Technologies, a leading provider of high‑performance networking hardware, sparking interest from multiple tech giants and analysts who see the deal as a potential boost for data‑center AI, big‑data analytics, and HPC capabilities.
