Summary: Google recently put a new cloud service into beta that enables developers to access NVIDIA GPU-based virtual machines from its cloud infrastructure platform. To get additional computing capacity and power, end users can simply attach the GPU-based VMs to existing Google Cloud machines. The capability is geared at developers working on processing-intensive workloads and enabling them to avoid procuring and deploying hardware. The idea is to replicate bare metal levels of performance with cloud-based infrastructure. Google is charging $0.70 per GPU per hour in the US and $0.77 in Europe and Asia-pacific. This is cheaper than its rivals. AWS’s basic service costs $0.90 per hour while Microsoft, which launched Nvidia-based Azure N-Series cloud instances late last year, charges $700 a month for a basic configuration.
↧