With the Tenzor language prototype, "speed and accuracy don't have to compete ... they can go hand in hand, together."

High-performance computations are needed for an increasing number of tasks - for example, image processing in neural networks or various machine learning programs.

It is believed that there are inevitable discrepancies between speed and reliability when performing such operations. According to this view, if speed is the main priority, then reliability is likely to suffer, or, conversely, the speed will decrease when you focus more on reliability.

However, a group of researchers, mostly from MIT, question this notion and claim that you can actually have both. Amanda Liu, a second-year doctoral student at MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL), says of the new programming language she wrote specifically for high-performance computing: "Speed and accuracy don't have to compete. Instead, they can work hand in hand on the programs we write”.

Liu warns that although ATL is promising, it is still a prototype tested in only a few small programs. "One of our main goals for the future is to improve the scale of the ATL so that it can be used for larger programs that we see in the real world."