Google Cloud has announced the general availability of its TPU virtual machines.
Tensor Processing Units (TPUs) are application-specific integrated circuits (ASICs) developed by Google that are used to accelerate machine learning workloads.
Cloud TPU lets you run your machine learning workloads on the cloud hosting giant's TPU acceleration hardware using the open source TensorFlow machine learning platform.
What can TPU virtual machines do for users?
Google says its user community has embraced virtual TPUs because they provide a better debugging experience and also allow for certain training configurations, including distributed reinforcement learning, which it says were not feasible with the UPT (Internet Access) node architecture. network) existing from Google.
Cloud TPUs are optimized for large-scale rating and recommendation workloads according to Google, citing how Snap was an early adopter of the capability.
Additionally, with the GA release of TPU VMs, Google is introducing a new TPU integration API, which it claims can speed up ML-based classification and recommendation workloads.
Google highlighted how many modern businesses rely on ranking and recommendation use cases, such as audio and video recommendations, product recommendations, and ad ranking.
The tech giant said TPUs can help companies implement a deep neural network-based approach to address the above use cases, which it says can be expensive and resource-intensive.
Google also claims that its TPU VMs offer several additional features over the existing TPU node architecture due to their local runtime configuration, since the input data pipeline can run directly on the TPU hosts, which enables organizations to save IT resources.
TPU VM GA Release also supports other major ML frameworks such as PyTorch and JAX.
Are you interested in implementing a virtual TPU? You can follow one of Google's quick start guides or tutorials.