Overview
Discover how to build and run AI workloads at scale while optimizing enterprise and cloud native applications in this 48-minute conference talk from Google Cloud Next 2024. Explore the latest innovations in supercomputing architecture, including AI Hypercomputer, TPUs, and GPUs. Learn about workload-optimized infrastructure best practices across compute, networking, and storage. Gain insights into increasing productivity and efficiency with AI-powered assistance using Gemini for Google Cloud. Hear from industry experts about real-world applications, performance optimization techniques, and the benefits of purpose-built infrastructure for various workloads. Dive into topics such as TPU V5, Nvidia GPUs, open-source software, block storage, dynamic workload scheduling, and Google Cloud Storage. Witness a demonstration of Gemini Cloud Assist and understand how it simplifies infrastructure operations.
Syllabus
Intro
Overview
Applications
Core Responsibilities
Our Strategy
Performance Optimized Infrastructure
TPU V5P
Nvidia GPUs
Open source software
Storage
Block storage
Networking
Dynamic workload scheduler
Salesforce
Workload optimized systems
Storage optimized systems
Google Axion
Google Cloud Storage
Gemini Cloud Assist
Demo
Wrapup
Taught by
Google Cloud Tech