TPU v4：面向机器学习的光学可重构超级计算机，具备嵌入硬件支持
TPU v4: An Optically Reconfigurable Supercomputer for Machine Learning with Hardware Support for Embeddings
AI总结：
本文介绍Google TPU v4超级计算机，利用OCS光学重构互连拓扑并以SparseCore加速嵌入，在4096芯片规模、性能功耗比及碳排放上显著优于同类系统。
摘要：
为响应机器学习（ML）模型的创新，生产工作负载发生了剧烈而迅速的变化。TPU v4是Google第五代领域专用架构（DSA），也是其第三代面向此类ML模型的超级计算机。光路交换机（OCS）动态重构其互连拓扑，以改善规模、可用性、利用率、模块化、部署、安全性、功耗和性能；如果需要，用户可以选择扭曲的3D环面拓扑。OCS及底层光学组件比Infiniband便宜得多、功耗更低且速度更快，其成本不到系统成本的5%，功耗不到系统功耗的3%。每个TPU v4都包含SparseCore，这是一种数据流处理器，可将依赖嵌入的模型加速5倍至7倍，而仅占用5%的芯片面积和功耗。TPU v4自2020年开始部署，其性能比TPU v3高2.1倍，性能/瓦提高2.7倍。TPU v4超级计算机的规模扩大4倍，达到4096个芯片，因此整体速度约快10倍，这与OCS的灵活性一起为大语言模型提供帮助。对于规模相近的系统，它比Graphcore IPU Bow快约4.3倍至4.5倍；与Nvidia A100相比，速度快1.2倍至1.7倍，功耗低1.3倍至1.9倍。在Google Cloud能源优化的仓库级计算机中，TPU v4与典型本地数据中心中的当代DSA相比，能耗约低3倍，产生的CO2e约少20倍。
英文摘要：
In response to innovations in machine learning (ML) models, production workloads changed radically and rapidly. TPU v4 is the fifth Google domain specific architecture (DSA) and its third supercomputer for such ML models. Optical circuit switches (OCSes) dynamically reconfigure its interconnect topology to improve scale, availability, utilization, modularity, deployment, security, power, and performance; users can pick a twisted 3D torus topology if desired. Much cheaper, lower power, and faster than Infiniband, OCSes and underlying optical components are <5% of system cost and <3% of system power. Each TPU v4 includes SparseCores, dataflow processors that accelerate models that rely on embeddings by 5x-7x yet use only 5% of die area and power. Deployed since 2020, TPU v4 outperforms TPU v3 by 2.1x and improves performance/Watt by 2.7x. The TPU v4 supercomputer is 4x larger at 4096 chips and thus ~10x faster overall, which along with OCS flexibility helps large language models. For similar sized systems, it is ~4.3x-4.5x faster than the Graphcore IPU Bow and is 1.2x-1.7x faster and uses 1.3x-1.9x less power than the Nvidia A100. TPU v4s inside the energy-optimized warehouse scale computers of Google Cloud use ~3x less energy and produce ~20x less CO2e than contemporary DSAs in a typical on-premise data center.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
