NVIDIA Tesla GPUs Come to Oracle Bare Metal Cloud

Over at the NVIDIA Blog, Kristin Bryson writes that the Oracle Bare Metal Cloud now offers Tesla P100 GPUs for technical computing.

Ian Buck from NVIDIA describes the advantages of Tesla P100 GPUs at Oracle OpenWorld.

Oracle Cloud customers can access NVIDIA Tesla P100 GPU accelerators, starting today. Oracle also announced plans to expand its cloud offerings to include Tesla V100 GPUs, the most powerful data center GPUs, based on our latest Volta architecture.

The move underscores growing demand for public-cloud access to our GPU computing platform from an increasingly wide set of enterprise users. Oracle’s massive customer base means that a broad range of businesses across many industries will have access to accelerated computing to harness the power of AI, accelerated analytics and high performance computing.

We’re working closely with NVIDIA to provide the next generation of accelerated computing to enterprises worldwide using our X7 compute architecture,” said Kash Ifikhar, vice president of Oracle Cloud Infrastructure. “This provides incredible flexibility to data scientists, engineers and researchers, allowing them to rent cutting-edge AI and HPC supercomputers by the hour to solve challenges of exceptional complexity.”

Oracle’s NVIDIA P100 offering provides its users two P100 GPUs with NVIDIA NVLink high-speed interconnect technology and can deliver 21 teraflops of single-precision performance per instance — the kind of performance required for deep learning training and inferencing, accelerated analytics and high performance computing.

Each P100 cloud instance can deliver the performance of up to 25 non-accelerated servers, dramatically saving money for HPC and AI workloads.

Accelerated computing is powering a revolution in AI, HPC and enterprise computing,” said Ian Buck, vice president of Accelerated Computing at NVIDIA. “Now with NVIDIA GPUs, the Oracle Cloud brings that computation power to its customers worldwide.”

Reducing Model Training from Weeks to Days

One of the first to access the new NVIDIA GPU offering from Oracle Cloud is Fluent.ai. Inc. — a member of NVIDIA’s Inception program for AI startups. Fluent.ai offers the world’s first acoustic-only speech recognition technology to OEMs wanting to make their consumer electronics voice enabled.

Fluent.ai’s proprietary neural network algorithms learn to understand meaning directly from a user’s speech. The result is a highly flexible, accessible and accurate voice-interface technology that performs robustly even in offline and noisy environments — in any language or with any accent.

Running on new NVIDIA GPU instances has significantly optimized the training of our deep learning models compared to the previous-generation hardware,” said Vikrant Tomar, chief technology officer at Fluent.ai. “This allows us to train more sophisticated speech recognition models while reducing the overall job time from weeks to days.”

And, cloud customers will be able to glean even more performance and cost-savings from our new NVIDIA V100 GPUs. With more than 120 teraflops of deep learning performance per GPU, a single Volta GPU offers the equivalent performance of 100 CPUs.

With the benefits of our GPU computing platform available to an even wider audience, Bryson says we can expect to be amazed by a new wave of solutions for problems not yet solved.

Sign up for our insideHPC Newsletter