Tech Giant, Google is extending its GPU offerings to Nvidia’s Volta deploying the GPU across a wider range of engines instead of gearing up to replace competitor GPUs with its own hardware. Google moves support several machine learning, general compute performance now available through cloud services.
As per the Google’s latest post, Nvidia V100 Tesla GPUs are available for Kubernetes Engines and the Compute Engine. Google’s Compute Engine is the cloud of servers that are used to run workloads while Kupernetes is typically used to scale containerized applications across the cloud.
In a single VM, an individual can stack eight Tesla V100 GPUs, 96vCPUs and 624GB of memory, which tends to be a bit more horsepower. The tech giant stated that Next-generation offers up to 300GB/s of GPU-to-GPU bandwidth. The company stated that this boosts performance in deep learning by up to 40 percent.
Google says, “NVIDIA V100s are available immediately in the following regions: us-west1, us-central1, and Europe-west4. Each V100 GPU is priced as low as $2.48 per hour for on-demand VMs and $1.24 per hour for Preemptible VMs.” Further, it added, “Like our other GPUs, the V100 is also billed by the second and Sustained Use Discounts apply.”
Tech giant recommends Nvidia’s P100 GPUs for workloads that need a balance between price and performance.