Nvidia Open-Sources CuFile APIs for Faster AI Storage Access

Nvidia has open-sourced its cuFile application programming interfaces and the vertical storage software stack beneath them. This move is part of an initiative called Storage-Next, which brings together more than 40 storage and flash suppliers to define how GPU-driven storage should operate.

CuFile allows GPUs to read from and write to storage directly instead of routing those requests through CPUs. This can reduce latency and improve performance for AI workloads that rely on large datasets and longer context windows.

The company has also outlined a framework called SCADA, which splits direct data access into two parts to address security risks. According to Nvidia, this approach can cut storage access times to microseconds.