By Stephen Nellis
SANTA CLARA, Calif. (Reuters) – Synopsys on Wednesday showed off a set of software tools designed to make it easier and faster to design cars, data centers and other large systems that rely on semiconductors.
Synopsys is one of the major players in making software for the design of the chips themselves, helping companies like Nvidia decide how to arrange hundreds of billions of transistors on tiny squares of silicon.
But with its $35 billion bid to buy engineering software company Ansys, Synopsys also aims to help customers design the products and systems where those chips will ultimately end up. At its annual developer conference Wednesday in Santa Clara, California, Synopsys CEO Sassine Ghazi described how some users are doing just that.
Tesla, for example, uses a virtual simulation of a custom chip to begin writing software for that chip and test how the software will control its cars long before any physical object is produced. In an interview, Sassine said other car buyers would follow suit, though he declined to name them.
“Tesla was a pioneer in looking at the car as a software-defined vehicle,” Ghazi told Reuters. “There are a number of European and Japanese OEMs going that route, and the Chinese are also rushing towards it.”
Synopsys also described how operators of massive data centers that run artificial intelligence systems like copilots or chatbots can simulate how that software will work on tens of thousands of chips — and how much heat the chips will release during the process, which then helps determine how much cooling equipment will be needed. needed.
“The speed at which artificial intelligence models can evolve and produce results is limited by the speed at which we can develop the supercomputer architecture and hardware to meet this new challenge,” Reynold D’Sa, corporate vice president of silicon, cloud hardware and infrastructure at Microsoft said in a statement. .
(Reporting by Stephen Nellis in Santa Clara, Calif.; Editing by Chizu Nomiyama)