Nvidia's Tesla GPU Technology Accessible Through Amazon Web Services Cloud
Nvidia on Monday said it will make its Tesla GPUs available to the enterprise space on-demand through the Amazon Web Service (AWS) cloud.
Nvidia's Tesla line of GPUs, which is now available through Amazon's Elastic Compute Cloud's offering, runs Nvidia's CUDA platform. The architecture utilizes compute-intensive visualization technology in the process of creating commercial products including design, testing and production.
"There are two typical use cases that have emerged for Cluster GPU Instances," said Andy Keane, general manager for the Tesla group, in an interview with CRN. "One is where people have an application, but they don't have the scale of cash or the facilities and would need one for their app. These are typically medium-sized businesses or startups. The second is for someone with a modest scale cluster. They won't have to scale it all up; instead, they can run it partly through the cloud."
The AWS cloud solution allows customers without sufficient resources to make the capital investment required to run CUDA. Keane said the extent of cost reduction from accessing GPUs through the cloud will vary depending on the scale of the business using the service.
"To build a data center-style or rack-style cluster, you need to equip a facility and make use of it, or it will sit there depreciating," he said. "Some businesses can justify having that physical facility. For the other folks, the Amazon facility allows them to keep a quarter of their capacity and get the rest from the cloud."
Launched in 2006, Amazon Web Services is a cloud platform that requisitions power, storage, and other services to an elastic IT infrastructure suite.
’With Amazon Cluster GPU Instances, our customers now have the power of high performance computing, the efficiency and speed of GPUs and the highly available, scalable and affordable cloud environment our customers have come to expect from AWS,’ said Peter De Santis, general manager of Amazon EC2, in a statement.
Nvidia says its Tesla GPUs are used for many kinds of high-performance computing, including structural analysis, financial simulations, video transcoding, and cancer research. Keane mentioned product design and biological simulation in particular as areas that could benefit from the additional capacity of Amazon's virtual facility in the prototyping stage of their product development.
Next: The Example of MatLab
Keane also singled out MatLab as a product which many have used for prototyping in the three years since Nvidia's Tesla GPUs have performed parallel processing.
The problem, Keane said, is that for many in product design, the application wasn't fast enough.
"Now you can use MatLab as a design tool running on a GPU, which usually moves faster than the CPu, and the cloud can help you scale it up," Keane said.