NVIDIA CUDA Tile is one of the most significant additions to NVIDIA CUDA programming and unlocks automatic access to tensor cores and other specialized hardware.

NVIDIA DevTalk serves as a vibrant community hub where developers can engage in discussions, seek assistance, and collaborate on projects involving NVIDIA hardware and software. Developers can tap into the collective expertise of the NVIDIA developer community, sharing insights, troubleshooting issues, and exploring best practices for GPU programming and AI development. Additionally, DevTalk provides a platform for developers to showcase their projects, receive feedback, and network with peers, fostering collaboration and knowledge exchange within the NVIDIA ecosystem.

NVIDIA Developer

cuTile.jl is a new Julia package that brings NVIDIA CUDA Tile-based GPU programming to Julia, mirroring the cuTile Python API while using idiomatic Julia syntax. It abstracts away thread/warp management, letting developers write kernels at the tile level while the compiler handles hardware mapping including tensor cores. Performance benchmarks on an RTX 5080 show near-parity with the Python implementation for common kernels like vector addition (99%), matrix transpose (98%), and matrix multiplication (100%). The package is experimental, open-source under JuliaGPU, and requires a Blackwell GPU with CUDA 13+ and Julia 1.11+.

cuTile.jl Brings NVIDIA CUDA Tile-Based Programming to Julia