NVIDIA/nccl

Optimized primitives for collective multi-GPU communication

C++CMakefileOther
This is stars and forks stats for /NVIDIA/nccl repository. As of 08 May, 2024 this repository has 2434 stars and 651 forks.

NCCL Optimized primitives for inter-GPU communication. Introduction NCCL (pronounced "Nickel") is a stand-alone library of standard communication routines for GPUs, implementing all-reduce, all-gather, reduce, broadcast, reduce-scatter, as well as any send/receive based communication pattern. It has been optimized to achieve high bandwidth on platforms using PCIe, NVLink, NVswitch, as well as networking using InfiniBand Verbs or TCP/IP sockets. NCCL supports an arbitrary number of GPUs installed...
Read on GithubGithub Stats Page
repotechsstarsweeklyforksweekly
thelfer/tfelC++CMakeMakefile770340
google/oboeC++JavaKotlin3.5k05360
Subhradeep10/FitnezzHTMLCSSJavaScript430550
toidicode/templateHTMLJavaScriptCSS75104050
ruby0x1/machinery_blog_archiveHTMLCSS1910500
mock-server/mockserverJavaHTMLJavaScript4.3k01k0
react-native-svg/react-native-svgJavaTypeScriptObjective-C++7k+111.1k0
OpenLineage/OpenLineageJavaPythonRust1.4k02120
pantheon-systems/WordPressPHPJavaScriptCSS15702290
p0dalirius/ApacheTomcatScannerPythonMakefile6700850