/
lectern.tv
Play15s30s1x0:00/--:--
DoneWatched
Expand
GPU MODE·Cris Cecka
1:44:03
0:00/1:44:03
Type below to lock the video timestamp, or press S to capture a slide.
PyCuTe: New Features and Abstractions for CUDA Tensors
An in-depth presentation on the latest release of PyCuTe, the reference implementation for CUDA Tensors, detailing new features, abstractions, and practical applications for GPU kernel development.
Join Cris Cecka and Mark Saroufim in GPU MODE to learn about the new release of PyCuTe, the reference implementation of CUDA Tensors (CuTe).
PyCuTe’s new features and abstractions - Examples and tutorials using PyCuTe for layout analysis - Using PyCuTe in CuTeDSL (and other Python DSLs) to develop GPU kernels