GoogleCloud Platform now has tensor processing units (TPUs) available for beta testing. Designed at Google for its own internal use, the TPUs are hardware accelerators that speed up the processing of machine learning workloads. Each TPU offers 180 teraflops of performance and 64 GB of memory. The chips can be linked together for high performance computing (HPC) purposes.
According to Google, the TPUs should greatly speed up the process of training machine learning models. “You can train several variants of the same model overnight on a fleet of Cloud TPUs and deploy the most accurate trained model in production the next day,” said a Google blog post.