ScaleWire:面向 LLM 系统的 Scale-Up 以太网端到端仿真框架
ScaleWire: An End-to-End Scale-Up Ethernet Simulation Framework for LLM Systems
搞 LLM 系统和硬件的人可以看看这篇:ScaleWire 用包级仿真加资源探索引擎,帮你评估不同网络接口设计。DeepSeek-R1 dispatch 场景下片上存储能差 4.5 倍,工具比穷举省 80% 以上的仿真量,做开放互连设计的团队用得上。
论文提出 ScaleWire,一个面向 LLM 训练与推理的 Scale-Up 以太网端到端仿真框架,核心是将以太网跨传输层、网络层、链路层的网络接口(NI)机制做包级仿真,再结合解析估算与仿真的资源探索引擎来规划片上资源。框架通过硬件实测和九项解析检查完成验证,并在四种 LLM 通信工作负载上评估了六种 NI 设计。
实验结果显示 NI 设计选择影响显著:在 DeepSeek-R1 的 dispatch 通信中,不同 NI 设计导致单个 XPU 片上存储预算出现 4.5 倍差异;资源探索引擎只需穷举搜索所需仿真量的 6-19%。作者还用 GH200 GPU 实测 kernel 时间的 vLLM prefill 轨迹,在完成时间约束下指导推理执行的资源配置。
对系统设计者而言,这项工作提供了在通信性能与片上资源成本之间做权衡的工具,对应的是 UALink、超级以太网等开放互连方案在 Scale-Up 场景落地的需求。
ScaleWire: An End-to-End Scale-Up Ethernet Simulation Framework for LLM Systems
LLM training and inference rely on scale-up networks for high-bandwidth, low-latency memory-semantic communication among accelerators (XPUs). Ethernet is emerging as an open interconnect for scale-up systems. In Ethernet systems, network-interface (NI) mechanisms across transport, network, and link layers interact, jointly shaping communication performance and on-chip resource requirements. We present ScaleWire, an end-to-end scale-up Ethernet simulation framework for LLM systems. It couples packet-level simulation of cross-layer NI mechanisms with a resource exploration engine that combines analytical estimates and simulation to size on-chip resources. We validate ScaleWire against hardware measurements and nine analytical checks, and evaluate six NI designs across four LLM communication workloads. For DeepSeek-R1 dispatch, NI design choices lead to a 4.5$\times$ difference in per-XPU on-chip storage budget. In grid-comparison experiments, the exploration engine uses 6-19% of the simulations required by exhaustive search. Using vLLM prefill traces with kernel times measured on GH200 GPUs, ScaleWire further guides resource provisioning under a completion-time constraint for inference execution. ScaleWire thus helps designers select NI mechanisms and provision on-chip resources to balance communication performance and resource cost.
本条情报归属 AITOP「论文」核心栏目,重点聚焦北美前沿实验室与硅谷科技巨头。在跨区域产业观察坐标中,该动态直接关联「AI 开发者工具链、Coding Agent 与工程化实践」研究主线,反映出当前技术演进与商业落地的最新风向。涉及核心实体与技术议题包括:ScaleWire、以太网、LLM训练、vLLM,读者可通过下方标签进入对应专题页查阅历史脉络与关联对比。
信源核验说明:本文原始信息发布自一手权威信源「arXiv: DeepSeek」(Tier-1 一手发布),具备高确定性参考价值。