Nov 14 (Reuters) – Silicon Valley artificial intelligence (AI) computing startup SambaNova Systems said on Monday it has delivered eight units of its latest AI system to the US Argonne National Laboratory, which is expanding its offering of AI to researchers.
Each of the systems features eight DataScale SN30 chips, launched in September, and triple the speed of the previous generation.
With AI work taking center stage in research, Argonne National Laboratory has been testing various AI chips and systems, Rick Stevens, associate lab manager for computing, told Reuters. environment and life sciences at the Argonne laboratory.
In addition to SambaNova, artificial intelligence systems from startups such as Cerebras Systems, Groq Inc, Graphcore and Habana Labs owned by Intel Corp (INTC.O) were tested.
“We’re working with new emerging AI hardware architectures, and we’re getting early hardware and then playing with it, using it on our science applications,” Stevens said.
“When we get to that point, we make internal decisions – ‘Is it useful enough, good enough and interesting enough for us to take the next step? “And so SambaNova took that step.”
Steven said the lab is evaluating AI hardware for its next supercomputer to see if SambaNova and others can be included. The lab, which is a Department of Energy national lab run by the University of Chicago, is building a supercomputer dubbed Aurora that will be an exascale two-speed machine — capable of performing two billion calculations in one second.
Stevens said the lab is in talks with Cerebras and Habana about scaling those systems as well. “What we’re trying to do is understand the strengths of these different systems,” he said.
SambaNova’s first-generation AI system has been used to predict how tumors respond to various drug combinations, improve the accuracy of weather forecasts and speed up fluid dynamics simulations, he said.
Reporting by Jane Lanhee Lee; Editing by Kenneth Maxwell
Our standards: The Thomson Reuters Trust Principles.