NVSHMEM Early Access
OpenSHMEM Communication on GPUs
NVSHMEM provides multi-GPU and multi-node communication primitives that can be initiated from the host or device, and called from within a CUDA kernel. NVSHMEM implements the OpenSHMEM standard for GPU memory, with extensions for improved performance on GPUs. Using NVSHMEM, applications automatically benefit from regular performance improvements and new GPU architectures.
NVSHMEM is currently available in an apply-for-access pre-release. Apply for access today!