Onramp for AI: AIRI Mini Lets Businesses Jumpstart Their Deep Learning Journey

NVIDIA and Pure Storage introduce an affordable way to build out AI infrastructure.
by Tony Paikeday

Sometimes you know exactly where you want to end up, but need guidance on how to take the first step.

For businesses building their deep learning systems, immediate considerations include getting started quickly and keeping upfront costs low, while getting the most GPU bang for their buck.

But these short-term considerations can steer businesses off course from where they want to end up longer term: with an enterprise-grade AI infrastructure that maximizes data scientist productivity, keeps their GPUs busy at any scale and minimizes cost.

AIRI Mini is the onramp to AI for enterprises. The result of a collaboration between NVIDIA and Pure Storage, AIRI Mini offers a scalable architecture for AI and deep learning, with a low-risk entry point that lets businesses stand up deep learning infrastructure quickly and cost-effectively.

With AIRI Mini, infrastructure design decisions made today won’t detour businesses from future destinations. That’s because it follows the same design best practices as found in its larger sibling, AIRI, by efficiently converging GPU computing, network fabric and storage into an optimized hardware and software stack.

And since it’s built on that same DNA, it inherits the same proven performance.

With either AIRI Mini or AIRI, the right network fabric and storage underpins each NVIDIA DGX-1 compute node within, so your environment can seamlessly scale. Simply add more DGX-1 nodes as deep learning training objectives demand it, along with more FlashBlade for additional storage.

This modular approach helps enterprises ensure that growing their GPU footprint reaps predictable increments in training performance without the penalty of inefficient intra-cluster communications or other overhead.

For a deeper look, check out our joint reference architecture, now updated for AIRI Mini.

If you’re at Pure Accelerate 2018, stop by NVIDIA Booth 501 to check out AIRI. Then continue the conversation at SAP SAPPHIRE, June 5-7, to take a closer look at how NVIDIA and Pure Storage are powering GPU-accelerated machine learning applications for business.