GPU Technology Conference, Munich: According to Nvidia founder and CEO, Jensen Huang, there are two dynamics controlling the computing industry today – the end of Moore’s law and software that can write itself, artificial intelligence, or AI.
The end of Moore’s law, which states that the density of a chip doubles every two years, while decreasing in price, has ended, as the reliance on the cpu has shifted to the graphics processor unit (gpu), Huang asserted today.
“Accelerated computing looks at the entire stack, algorithms, software and processor,” he said. “We can study where bottlenecks are. New software systems make the application go faster, not just the chip,” he explained.
“Nvidia is a one trick pony,” he confessed. “We are investing in accelerated computing and selecting markets and problems where computing can perform and fill a gap; the only way to solve a problem, in retail, telecommunications, automotive, is to re-engineer the stack, from processor chips, algorithms, software – from the top to the bottom.”
Accelerated computing is liberating, said Huang. “Let’s say you have an airplane that has to deliver a package. It takes 12 hours to deliver it. Instead of making the plane go faster, concentrate on how to deliver the package faster, look at 3D printing at the destination,” he explained. Applications acceleration is not just to make the chip go faster, it is to deliver the goal faster.”
At this week’s conference, Huang introduced Rapids, a suite of software libraries for AI. The open source gpu acceleration platform is designed for large scale data analytics and machine learning.
At the same time, he announced that it had been trialled by Walmart, tracking stock and logistics, and is supported by Hewlett Packard Enterprise, IBM and Oracle, and open source partners, such as Databricks and Anaconda. It adds gpu acceleration to popular open source Python data science tools and will be integrated into Apache Spark, described as the leading open source framework for analytics and data science.