Users of Google's Compute Engine and Kubernetes Engine are now able to use Nvidia Tesla V100 GPUs as a beta release, and at the same time, the search giant has announced the general availability of Tesla P100 GPUs.
The V100 GPUs are available in us-west1, us-central1 and europe-west4 regions, with pricing starting from $2.48 per hour for on-demand VMs, and $1.24 for preemptible VMs. The cost of the P100 is now $1.46 per hour on-demand, and $0.73 hourly for preemptible VMs.
In a blog post, Google said single VMs can now have up to eight V100 GPUs, 96 vCPU, and 624GB of memory. The use of V100 within a virtual machine is limited to either 1 or 8 GPUs, with 2 and 4 configurations in beta testing, and for the P100, a single VM can have only 1, 2, or 4 GPUs.
Unlike the GPU announcement, there was no pricing available for using dedicated Isilon instances.
Last week Google announced a number of managed database improvements, including commit timestamps for Cloud Spanner, Cloud Bigtable replication in beta, Cloud Memorystore for Redis in beta, and the general availability of Cloud SQL for PostgreSQL.