Engineers at Google have tasked artificial intelligence with designing faster and more efficient processors and then used its chip designs to develop the next generation of specialised computers that run the very same type of AI algorithms. Google operates at such a large scale that it designs its computer chips rather than buying commercial products.
This enables it to optimise the chips to run its software, but the process is time-consuming and expensive. A custom chip takes two to three years to develop.The first stage of chip design is a process called floor planning, which includes taking the finalised circuit diagram of a new chip and arranging the millions of components into an efficient layout for manufacturing.The functional design of the chip is complete. The layout can have a huge effect on speed and power consumption. For chips in smartphones, the priority may be to cut power consumption to increase battery life, but for a data centre, it may be more important to maximise speed.
Anna Goldie at Google said that floorplanning was a highly manual and time-consuming task. Teams would split larger chips into blocks and work on parts in parallel, fiddling around to find small refinements, she says. Now Goldie and her colleagues have created software that turns the floorplanning problem into a task for a neural network.
The new software treats a blank chip and its millions of components as a complex jigsaw with a vast amount of possible solutions. The aim is to optimise whatever parameters the engineers decide are most important, while also placing all the components and connections between them accurately.The software began by developing solutions at random that were tested for performance and efficiency by a separate algorithm and then fed back to the first one. In this way, it learned what strategies were effective and built upon past successes. Goldie also said that it started randomly and gets really bad placements, but after thousands of iterations it becomes extremely good and fast.