Quantum Applications Quantum Hardware

Google’s TPU’s to simulate Quantum Workloads with Floq API

March 12, 2021

Google has built a processing device named the TPU or Tensor Processing Unit. You have certainly used a CPU and most likely a GPU (Graphical Processing Unit), but the TPU presents a new way of processing data. Since 2015 Google has used TPU’s in their own development. The Tensor Processing Unit (TPU) is specific hardware developed by Google for neural network machine learning, especially tailored to Google’s Machine Learning software named Tensorflow. But now the team aims to repurpose the technology to allow Quantum workloads to be run using a new API named Floq.

Google’s Tensorflow for Quantum – already one year old

Companies such as Graphcore (based in the UK) are also working on alternative processing technologies named the IPU and there is a degree of excitement in the computing community about advanced processing techniques. After all the rise of the GPU has enabled massive acceleration in the deep learning and machine learning spaces. Google was one of the first large technology companies to release a major machine learning tool: Tensorflow. Tensorflow from Google has continuously evolved and developed and now sports Quantum integration for machine learning. That isn’t to confuse Google’s native Quantum language Cirq (which has just announced its 0.10 release).

A recent report from VentureBeat claims that Google, or Alphabet, the parent company of Google will be utilising TPU’s in the pursuit of simulating Quantum Workflows. The report claims that the technology giant will be releasing a new API named Floq to facilitate the running of quantum workflows on the TPU.

The Sandbox at Alphabet team repurposed TPUs (Tensor Processing Units), to accelerate simulations in the cloud so that researchers and developers can use tools such like TensorFlow Quantum and PennyLane to create quantum models. Floq the API debuted at QHack 2021, organised in part by Guillaume Verdon, who is not only a Quantum scientist but has co-founded / co-developed Tensorflow. The teams at QHack worked on a variety of challenges related to Quantum Machine Learning (QML) back in February 2021 competing for prizes and the fame of winning the attention of some of the most well known Quantum researchers.

As yet we don’t know when the Floq tools will be available or how the public can get involved with utilising the technology stack. But we look forward to hearing great things from Alphabet very soon!