Header Ads

Nvidia Tesla K80 GPUs added to Google Compute Engine and Cloud Machine Learning



Google has added beta support for Nvidia Tesla K80 GPUs to allow Cloud Platform customers to get extra computational power for deep learning tasks.

Bolster for the Nvidia GPU-fueled virtual machines will be accessible in three Google Cloud Platform locales - us-east1, asia-east1 and europe-west1 - with support for making GPU virtual machines utilizing the Cloud Console being coming one week from now. 

Clients can now connect up to eight GPUs (four Nvidia Tesla K80 sheets) to any custom Google Compute Engine virtual machine. Each GPU has 2,496 stream processors and 12 GB of GDDR5 memory 

These GPUs can be utilized to quicken an assortment of registering and examination workloads, from fundamental errands, for example, video and picture transcoding, to more confounded procedures, for example, seismic investigation, atomic displaying, back, liquid progression and representation.


The GPUs will bolster an assortment of machine learning and profound learning structures, for example, TensorFlow, Theano, Torch, MXNet and Caffe, and also Nvidia's own particular CUDA stage. 

Evaluating is focused, with each K80 GPU appended to a VM costing just $0.700 every hour per GPU in the US, and $0.770 every hour per GPU in Asia and Europe. Clients pay for what they use, with no venture costs for equipment.

No comments:

Powered by Blogger.