Protocol

GPUDirect Storage

GDS

Lab ready Full lab

GPUDirect Storage (GDS) provides a direct DMA path between NVMe and GPU memory, skipping host bounce buffers. Filesystem and driver support plus correctly registered buffers are required. It accelerates data ingest for training and high-throughput inference feature pipelines.

Domains

Tags

gpunvmezero-copystorage

Sources

  • NVIDIA Magnum IO GPUDirect Storage documentation

Related protocols

NVMe

Lab ready

NVM Express — PCIe register and queue interface for SSDs with submission/completion queues, PRP/SGL data pointers, and admin commands. Pa…

storageai

NVMe over Fabrics

Lab ready

NVMe over Fabrics extends NVMe queues across RDMA, TCP, or Fibre Channel. Capsules map submission/completion queue entries over the fabri…

storageai

GPUDirect RDMA

Lab ready

NIC DMA path that reads and writes GPU memory directly, bypassing host bounce buffers for RDMA verbs. Requires peer-to-peer BAR mapping, …

ai

dma-buf

Lab ready

Linux buffer-sharing primitive for zero-copy exchange across drivers such as GPU, NIC, and media. Exporters and importers negotiate acces…

aistorage