That program was originally able to optimize the layout of circuits in the floor plan of the chip, the way that the two-dimensional area is used. The program, called DSO.ai, was first introduced a year ago in May. The conference is being held virtually this year.Īlso: AI on the bench: Cadence offers machine learning to smooth chip designĭe Geus was explaining to ZDNet an expansion of AI technology in the company's software that has been several years in the making. "A machine will optimize everything, everything." That means a machine will take risks, narrowing the margin for error beyond what humans consider acceptable.ĭe Geus spoke with ZDNet in advance of a keynote address he is giving Monday at the annual Hot Chips computer chip conference for advanced computing. "Margin is essentially a risk calculation, and that is a sheer impossible thing to do for a human," explains Aart de Geus, chief executive of chip-design software maker Synopsys. A human wants to leave as wide a margin for error as possible. Designers will leave a margin for error when they place circuits on a chip, to anticipate errors in manufacturing that could, for example, throw off the timing of a signal making its way around the chip. "A machine will optimize everything, everything."Īrtificial intelligence is increasingly being utilized in the design of semiconductors, and one of its advantages is that AI technology will explore design trade-offs that humans would refuse to even consider.Ĭonsider the concept of margin, for example. "Margin is essentially a risk calculation, and that is a sheer impossible thing to do for a human." explains Aart de Geus, chief executive of chip-design software maker Synopsys.
0 kommentar(er)
