Google Cloud First to Offer Intel Skylake, Announces GPUs for CE and Cloud ML

Google Cloud Platform (GCP) supports as many as 8 GPUs attached to custom VMs. Becomes, the first cloud provider to offer next generation Intel Xeon processor, codenamed Skylake.

Share online:

Google Cloud Platform becomes the first ever cloud provider to offer the next generation Intel Xeon processor, codenamed "Skylake," available now in five GCP regions including: Western US, Eastern US, Central US, Western Europe and Eastern Asia Pacific.

With Skylake processors, "GCP customers are the first to benefit from the next level of performance," Google writes.

Those who are not aware, "Skylake includes Intel Advanced Vector Extensions (AVX-512), which doubles the floating-point performance for the heaviest calculations. Thus making it ideal for scientific modeling, genomic research, 3D rendering, data analytics and engineering simulations."

Google notes, that for the Compute Engine's family of VMs that include: standard, highmem, highcpu and Custom Machine Types —they've optimized Skylake to help bring the next generation of high performance compute instances to everyone.

With the much anticipated public beta of NVIDIA Tesla K80 —GPUs are also now available for Google Compute Engine and Cloud Machine Learning (Cloud ML), enabling customers to spin up NVIDIA GPU-based VMs in three GCP regions: us-east1, asia-east1 and europe-west1, using the gcloud command-line tool.

While support for creating GPU VMs using Cloud Console will appear next week, and enable those need extra computational power to add up to 8 GPUs (4 K80 boards) to any custom GCE VM, providing bare-metal performance.

"Each NVIDIA GPU in a K80 has 2,496 stream processors with 12 GB of GDDR5 memory. You can shape your instances for optimal performance by flexibly attaching 1, 2, 4 or 8 NVIDIA GPUs to custom machine shapes," wrties Google.

These instances support popular machine learning and deep learning frameworks such as TensorFlow, Theano, Torch, MXNet and Caffe, as well as NVIDIA's popular CUDA software for building GPU-accelerated applications.

The GPUs are billed per minute (10 minute minimum), see the pricing chart below:

Google Cloud Engine GPUs Pricing Chart
Google Cloud Engine GPUs Pricing Chart

Google also highlighted three steps to help you secure Elasticsearch on its Cloud Platform (GCP), which is an open source search engine built on top of Lucene that is commonly used for internal site searches and analytics.

"Elasticsearch can be attacked using the following instance: "poisoning the index, exploiting unauthorized API access and exfiltrating sensitive data." To mitigage these attacks, Google recommends:

  • Lock down your access policy. Since, Elasticsearch relies on external access management, thus, as an important step is to lock down the access policy.
  • Don't index sensitive data. Once the access policy is in place, as a best practice "carefully filter out personally identifiable information (PII), cardholder data or other sensitive information to prevent it from leaking."
  • Handle unfiltered content safely. Index poisoning occurs when unfiltered malicious content is ingested by Elasticsearch. If you index user-generated content (UGC), be sure to properly filter it before storing it, explains Google.