Electronic design automation (EDA) is full of large, intricate problems. Figuring out the best way to arrange transistors on a chip, for example, becomes exponentially more complex as the number of transistors increases. Computer scientists have made great strides in developing algorithms that can solve many of these problems, but a team of researchers at the University of Michigan, in Ann Arbor, believes that the industry could benefit from a different resource: human intuition.
“These kinds of problems are difficult for computers to solve. We started by thinking, ‘How can humans help electronic design automation?’ ” says Valeria Bertacco, an associate professor in computer science and engineering. She and Andrew DeOrio, a doctoral student, have developed an online game that challenges players to take on a type of problem common in design automation. They presented their idea of human-assisted problem solving today at the Design Automation Conference, in San Francisco.
In large, complex EDA problems, there are initially millions of possible paths to a solution. It’s similar to a maze: At the beginning, you have to pick an initial path to explore and see where it leads. The problem for designers is that the number of solution paths increases exponentially with the number of variables. Even the best algorithms can get tripped up if they start down a search path with no possible solution.
Source: spectrum.ieee.org
No comments:
Post a Comment