A cheap, non-toxic material, commonly found in household paint, could hold the key to one of machine learning’s biggest issues. A new scientific paper published by a team from Sandia National Laboratories, alongside collaborators from the University of Michigan, has discovered that titanium oxide, often found in paints and varnishes, could dramatically improve the energy efficiency of computer chips.
By heating a computer chip coated with titanium oxide above 150 degrees Celsius, it is possible to remove some of the oxygen molecules, creating what is known as oxygen vacancies. This makes the material electrically conductive and, therefore, able to store information.
Crucially, the discovery could fundamentally change how computers store and process data, resulting in huge energy savings.
Currently, computers store data in one place before transferring it somewhere else for processing. This constant transferral of information creates a huge strain in terms of energy consumption and processing power. By cutting down on this unnecessary power loss, oxygen vacancies could open up opportunities to explore more energy-intensive applications.
“What we’ve done is make the processing and the storage at the same place,” Yiyang Li, the paper’s lead author explained. “What’s new is that we’ve been able to do it in a predictable and repeatable manner. If we are trying to do machine learning, that takes a lot of energy because you are moving it back and forth and one of the barriers to realizing machine learning is power consumption.”
The discovery of this new use for titanium oxide has potential applications for autonomous vehicles, image processing and voice recognition. These are, however, early days. Researchers still need to refine the process and look at ways to scale it up before oxygen vacancies will be employed in consumer or business technology.
Via HPC Wire