arXiv cs.AI· Sajy Khashab, Albert Gran Alcoz, Alon Gal, Jacky Romano, Rani Abboud, Yonatan Piasetzky, Lior Maman, Amit Nishry, Barak Gafni, Omer Shabtai, Matty Kadosh, Dror Goldenberg, Gilad Shainer, Mark Silberstein·· 12 小时前AI 评分38
NVIDIA Spectrum-X：面向大规模 AI 训练的硬件加速多平面网络
NVIDIA Spectrum-X: Hardware-Accelerated Multiplane Networking for Large-Scale AI Training
AI 导读
NVIDIA 提出 Spectrum-X 多平面架构，用拓扑并行替代层级深度，并在 NIC 与交换机中引入硬件加速负载均衡，以微秒级响应 AI 训练的动态网络变化。评测显示其达到理论线速的 98% 且延迟低抖动，10% 链路故障下延迟仅增加 7%，并保持跨租户隔离与对 LLM 训练中链路抖动的快速反应。
正文 · 原文
Authors:Sajy Khashab, Albert Gran Alcoz, Alon Gal, Jacky Romano, Rani Abboud, Yonatan Piasetzky, Lior Maman, Amit Nishry, Barak Gafni, Omer Shabtai, Matty Kadosh, Dror Goldenberg, Gilad Shainer, Mark Silberstein
Abstract:As distributed model training scales to span hundreds of thousands of GPUs, scale-out networks face unprecedented performance and efficiency demands. NVIDIA Spectrum-X Ethernet has been designed from the ground up to achieve predictable and stable network performance with high utilization and low latency. This paper presents the Spectrum-X multiplane architecture, which replaces hierarchical depth with topological parallelism, and introduces hardware-accelerated load balancing in NICs and switches as the key architectural approach to provide fast reaction to highly dynamic network conditions at the microsecond timescales that AI training workloads demand. We describe the motivation, design principles, evaluation methodology and performance on state-of-the-art benchmarks, as well as the lessons we learned from deploying and debugging Spectrum-X networks in large-scale systems. Our evaluation highlights production-grade AI infrastructure performance across four core dimensions: 98% of the theoretical line rate with low jitter-free latency; strong cross-tenant isolation for concurrent workloads; robust, capacity-proportional bisection bandwidth and 7% latency increase for 10% fabric link failures; and rapid reaction to host and fabric link flaps during LLM training workloads.
| Subjects: | Networking and Internet Architecture (cs.NI); Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC) |
| Cite as: | arXiv:2605.21187 [cs.NI] |
| (or arXiv:2605.21187v2 [cs.NI] for this version) | |
| https://doi.org/10.48550/arXiv.2605.21187 arXiv-issued DOI via DataCite |
Submission history
From: Sajy Khashab [view email]
[v1]
Wed, 20 May 2026 13:52:47 UTC (9,496 KB)
[v2]
Tue, 20 Oct 2026 08:25:21 UTC (4,657 KB)
来源：arXiv cs.AI · arxiv.org