Nvidia and Menlo Micro partnership speeds up AI chip testing
Nvidia and Menlo Micro have announced a collaboration that dramatically accelerates testing of Nvidia’s AI chips, helping to ease a critical production bottleneck.
AI chips require rigorous testing on specialized circuit boards to ensure they meet performance targets like speed and functionality. However, much of the testing hardware has remained decades-old, making it challenging to handle the high power consumption and extreme speeds of modern AI chips, News.Az reports, citing Reuters.
Menlo Micro, a startup spun out of GE in 2016 and backed by $227.5 million in funding from Corning and Tony Fadell’s venture fund, provides innovative switching chips that act like micro-scale metal light switches. These chips enhance the performance of Nvidia’s testing boards.
According to a joint research paper, the new system allows Nvidia to test its graphics processing units (GPUs) 30% to 90% faster depending on the test type. Russ Garcia, Menlo Micro’s CEO, said the technology is also being adopted by other major chipmakers.
"The bottom line is, if you don’t validate the GPUs before they enter data centers, you risk errors and operational issues," Garcia explained. "This is the only way to validate these things at speed."
With Nvidia reporting earnings Wednesday and analysts expecting 56% sales growth to $56.9 billion, the faster testing process comes as the AI chipmaker navigates surging demand for its products.
Similar news
