V
主页
京东 11.11 红包
How Chiplets and Open Interconnect Accelerate Cloud-Optimized AI Silicon
发布人
https://www.youtube.com/watch?v=6F9r4uK_Cog 演讲人:Mark Kuemerle(Marvell)| Ramin Farjadrad(Eliyan) 当今人工智能的发展给超大规模计算基础设施带来了压力,它们所依赖的传统半导体扩展技术已开始达到自然极限。这加快了对硅设计创新的需求,以提供性能、功耗和空间效率方面的飞跃,跟上人工智能革命的速度。即使是最先进的工艺节点,AI/ML 加速器和高性能计算(HPC)芯片也会受到微粒限制,因此芯片有望将摩尔定律引向更模块化、更垂直的方向,从而推动高性能 AI 和计算的发展。本专家小组由顶级行业分析师主持,将探讨芯片驱动设计的基本技术构件,以及芯片到芯片互连的开放标准替代方案,包括 NVLink、BoW 和 UCIe。
打开封面
下载高清视频
观看高清视频
视频下载器
Optical Interconnect:Pathways to an Open Infrastructure for AI
大规模 AI 集群网络技术 _ NVIDIA On-Demand
The Ultimate Chiplet Interconnect - Eliyan
Intro to Advanced Packaging
Advanced package tech. for chiplet adoption and memory integration in AI/HPC
UCIe™ (Universal Chiplet Interconnect Express™)
Navigating the AI ASIC Frontier Trends, Challenges, and Breakthroughs
Filling UEFI FW Gaps in the Cloud
Broadcom Accelerates PCIe/CXL Roadmap to Enable the Open AI Ecosystem
Optimized Storage from Supermicro and Solidigm to Accelerate Your Al Data Pipeli
Broadcom-EN141 Webinar_ RoCE Introduction
Flexible Data Placement Open Source Ecosystem
Extending PCIe Connectivity with Optics
Debug ability and Debug Practices of AI GPU Systems
SDC2022 – Persistent Memories Without Optane, Where Would We Be
UCIe : An Open Standard for Innovations at the Package Level
Open Infrastructure Platforms to usher in age of GenAI
Google TPU & other in-house AI Chips
AI Era Event - Marvell AI Day2024
Lenovo Blackwell-Powered ThinkSystem AI Portfolio Built
OCP China 2024 数据中心基础设施论坛
InfiniBand the Network that Accelerates HPC and AI Environments - Try it yoursel
Intel Xeon 6 with P-cores and Gaudi 3 AI Accelerators
Supercharging AI Innovation Through Purpose-Built Connectivity - Astera Labs
Storage Requirements for AI
Leveraging UEC For Next Generation AI Networks
OCP China Day 2023-主论坛
Hardware Management for Liquid Cooling: Enabling Efficient Operations
Optics in AI Clusters - Meta Perspective
OCP NIC in 2023 800G, PCIe Gen6 & Test Integration
Datacenter Physical Automation at Meta
Benchmarking Storage with AI Workloads
SERDES from 224 and beyond
OCP Rack & Power Project call (Aug 14, 2024)
Memory Corrected Error profiling via Linux EDAC Driver within large scale cloud
Applying AI/ML Methodologies to Categorize Storage Workloads
OCP China 2024 -- 开放计算生态论坛
Broadcom-Ethernet NIC Webinar Series (EN101)
Broadcom EN145 Webinar_ Multi-Host, Multi-Root and SR-IOV
Netdev 0x17 - [Keynote] Machine Learning Workloads And Networking