Optimized primitives for collective multi-GPU communication
updated at May 12, 2024, 6:30 a.m.
NVIDIA® TensorRT™ is an SDK for high-performance deep learning inference on NVIDIA GPUs. This repository contains the open source components of TensorRT.
updated at May 12, 2024, 1:26 a.m.
[ARCHIVED] The C++ parallel algorithms library. See https://github.com/NVIDIA/cccl
updated at May 11, 2024, 9:59 p.m.
An efficient C++17 GPU numerical computing library with Python-like syntax
updated at May 11, 2024, 3:30 a.m.
[ARCHIVED] The C++ Standard Library for your entire system. See https://github.com/NVIDIA/cccl
updated at May 8, 2024, 12:26 p.m.