← Back
NVIDIA Blog

As AI Increases Demands on Memory, Storage Steps Up

5 min read
#inference#compute
Level:Advanced
For:AI Engineers
TL;DR

The increasing demands of AI on memory and storage have led to the need for more efficient and secure storage architectures, with NVIDIA unveiling new storage advancements at the Future of Memory and Storage (FMS) conference. The NVIDIA Vera CPU delivers up to 3.21x higher throughput than an x86 CPU in a two-stage compression and encryption pipeline, enabling storage platforms to absorb AI data more efficiently. The open sourcing of NVIDIA cuFile APIs enables interoperability for storage solutions, allowing GPUs to read from and write to storage directly. This development has significant implications for engineers building AI systems, as it enables faster and more secure access to data and storage.

⚡ Key Takeaways

  • The NVIDIA Vera CPU delivers up to 3.21x higher throughput than an x86 CPU in a two-stage compression and encryption pipeline.
  • The cuFile API enables securely accessing data from storage in just microseconds, using hundreds of thousands of GPU threads and fast high-bandwidth memory.
  • The open sourcing of cuFile APIs enables interoperability between GPUs and data, providing a security-first storage stack based on Linux best practices.
  • The cuFile API is a component of NVIDIA GPUDirect Storage, which allows GPUs to read from and write to storage directly.
  • The Open Secure AI Alliance is a new initiative that aims to provide a foundation for secure AI development, with Google, Intel, NVIDIA, and Meta as inaugural maintainers.
💡 Why It Matters

The increasing demands of AI on memory and storage require more efficient and secure storage architectures, and the developments announced by NVIDIA have significant implications for engineers building AI systems. The open sourcing of cuFile APIs enables faster and more secure access to data and storage, which is critical for powering preventive and detective cybersecurity measures.

✅ Practical Steps

  1. Use the NVIDIA cuFile API to enable securely accessing data from storage in just microseconds.
  2. Integrate the cuFile API with NVIDIA GPUDirect Storage to allow GPUs to read from and write to storage directly.
  3. Explore the Open Secure AI Alliance initiative to learn more about secure AI development and the use of open technologies.

Want the full story? Read the original article.

Read on NVIDIA Blog

More like this

GLM-5.3 is here with advanced cyber capabilities — and reportedly already found a 'serious vulnerability' in Cursor

VentureBeat AI#llm

Universitas Gadjah Mada, Indosat and NVIDIA Open Indonesia’s First University AI Center to Develop Local AI Talent

NVIDIA Blog#nvidia

A decade of mathematical certainty: Reflections on the Automated Reasoning Group

Amazon Science#inference

With a feel for physics, AI models simulate a wider range of real-world scenarios

MIT News AI#llm

EXPLORE AI NEWS

Daily hand-picked stories on LLMs, RAG, agents and production AI — curated for engineers who ship.

BROWSE NEWS

GET THE WEEKLY DIGEST

Join engineers getting the Monday signal-over-noise AI breakdown. No spam, unsubscribe anytime.

LEARN AI ENGINEERING

Curated courses, research papers, repos and tutorials built for engineers leveling up in AI.

START LEARNING