Warp: A High-performance Python Framework for GPU Simulation and Graphics
, Principal Engineer, NVIDIA
高度评价
I'll introduce Warp, a runtime framework that allows running Python kernel code in CUDA. Warp aims to give users the productivity of developing in Python, with the performance of handwritten C++ and CUDA. I'll cover the code-generation techniques used to convert the Python AST to CUDA, and talk about some of the built-in functionality that allows implementing complex physical models such as cloth, FEM, and rigid body in a few hundred lines of Python. I'll also cover the integration of Warp into Omniverse, and show how it can be used to construct real-time and differentiable simulators for visual effects, machine learning, and robotics.
活动: GTC Digital Spring
日期: March 2022
话题: Accelerated Computing & Dev Tools - Libraries / Runtimes