Google has long been talking about offering support for high-end graphics processing units. The company appears to be aiming towards developers who need to power complex structures such as machine learning frameworks. Google’s recent announcement is actually built on a three months earlier statement in which the company announced that it plans to launch support for high end graphics processing units (GPUs).
With the onset of 2017, Google has finally brought a performance boost to its cloud platform with GPUs. It has been made available to developers for machine learning and other specialized workloads. The much anticipated public beta of NVIDIA Tesla K80 GPUs will help developers to attach up to eight of these to any custom Compute Engine machine.
Developers can now spin up NVIDIA GPU-based VMs in three regions i.e. us-east1, asia east1, and Europe-west1 by applying the gcloud command-line tool. Each of the NVIDIA Tesla K80 GPUs features 2,496 of stream processors with 12 GB of GDDR5 memory.
While using deep learning framework such as Tensorflow, Torch, MXNet of Caffee, developers can never have too much compute power with complex stimulations. This feature is clearly for developers who need to spin up clusters of high-end machines to power their machine learning platforms on daily basis.
Google’s Cloud machine learning service is incorporated to the Cloud GPUs as well as its database and storage platforms. In the European and Asian data centers, the price per GPU is $0.77 per hour where as it is $0.70 in the US. A Tesla K80 accelerator along with two cores as well as a 24 GB of Ram is quite much to offer with a bit high price.
Google will be hosting its Cloud Next conference in San Francisco and the audience can expect to hear more about the company’s plans for expanding its machine learning services to even more developers.