BKP/helion
A Python-embedded DSL that makes it easy to write fast, scalable ML kernels with minimal boilerplate.
A Python-embedded DSL that makes it easy to write fast, scalable ML kernels with minimal boilerplate.
Modular RDMA Interface
A framework to understand RDMA
ROC_SHMEM intra-kernel networking runtime for AMD dGPUs on the ROCm platform.
CUDA Templates for Linear Algebra Subroutines
TransferBench is a utility capable of benchmarking simultaneous copies between user-specified devices (CPUs/GPUs)
OpenSHMEM Application Programming Interface
Dockerfiles for the various software layers defined in the Radeon Open Compute Platform
MSCCL++: A GPU-driven communication stack for scalable AI applications
This repository contains Kokkos Remote Spaces, which implements distributed shared memory support for Kokkos.
The C++ parallel algorithms library.