Nvidia Launches Storage-Next Initiative for AI
Nvidia has launched Storage-Next, a new industry initiative with over 40 flash and storage vendors, aiming to optimize storage for AI computing. The company has also open-sourced its cuFile APIs and storage stack, and unveiled its SCADA framework, which moves storage control to the GPU.
Key points
- Nvidia has launched Storage-Next, an industry initiative with over 40 flash and storage vendors, to optimize storage for AI computing.
- The company has open-sourced its cuFile APIs and storage stack, making them available on GitHub.
- Nvidia has unveiled its SCADA framework, which moves storage control to the GPU, allowing parallel GPUs to pull data directly from storage.
- The SCADA framework aims to address the 512-byte problem in storage, where current SSD controllers spend the same effort on small and large data sizes.
- Nvidia's cuFile APIs have been available since 2021, but the company has now open-sourced the underlying storage stack.
- The Storage-Next initiative is seen as a key step in the development of AI computing, with experts arguing that storage infrastructure is becoming increasingly important.
Nvidia Launches Storage-Next Initiative for AI
Nvidia has made a significant move in the world of artificial intelligence (AI) computing with the launch of Storage-Next, a new industry initiative aimed at optimizing storage for AI workloads. The company has also open-sourced its cuFile APIs and storage stack, and unveiled its SCADA framework, which moves storage control to the GPU.
The Storage-Next initiative brings together over 40 flash and storage vendors, with the goal of developing more efficient storage solutions for AI computing. This is seen as a key step in the development of AI, with experts arguing that storage infrastructure is becoming increasingly important.
Nvidia's cuFile APIs have been available since 2021, but the company has now open-sourced the underlying storage stack. This move is expected to accelerate the development of more efficient storage solutions for AI workloads.
The SCADA framework is a key component of the Storage-Next initiative, and is designed to move storage control to the GPU. This allows parallel GPUs to pull data directly from storage, reducing the need for CPU involvement and improving overall system performance.
The 512-byte problem is a significant challenge in storage, where current SSD controllers spend the same effort on small and large data sizes. The SCADA framework aims to address this problem, and is seen as a key step in the development of more efficient storage solutions for AI workloads.
Sources
The WireByte editorial team synthesises technology news from multiple primary sources, verifies the facts, and links every source. Articles are produced with AI assistance and reviewed under our editorial policy.