Close

Presentation

An Extensible, Data-oriented Architecture for High-performance, Many-world Simulation
SessionCloud Rendering: Your GPU Is Somewhere Else
DescriptionTraining agents to learn skills in simulated environments is an increasingly important workload at the intersection of AI and graphics. The Madrona framework allows users to build high-performance simulators for novel tasks by leveraging entity-component-system (ECS) interfaces that map task logic onto parallel GPU execution across thousands of environments.
Authors
Brennan Shacklett
Stanford University
Luc Guy Rosenzweig
Stanford University
Bidipta Sarkar
Stanford University
Zhiqiang Xie
Stanford University
Andrew Szot
Georgia Institute of Technology
Erik Wijmans
Georgia Institute of Technology
Vladlen Koltun
Vladlen Koltun
Dhruv Batra
Georgia Institute of Technology
Kayvon Fatahalian
Stanford University
Event Type
Technical Paper
TimeWednesday, 9 August 20239:55am - 10:06am PDT
LocationPetree Hall D
Interest Areas
Research & Education
Recordings
Livestreamed
Recorded
Keywords
Real-Time
Rendering
Registration Categories
FC
FCS
V
VS
EFC
Next PresentationNext Presentation
Cloud Rendering: Your GPU Is Somewhere Else - Q&A and Interactive Discussion