Intel also wants to fight Covid-19 with its Loihi neural chip
Intel Promotes Pohoiki Springs Neural Chip Network Using Loihi Chips and FPGAs to Combat Covid-19.
Intel said yesterday Wednesday (March 18, 2020) that it had completed its Pohoiki Springs platform , the name of its network of 768 Loihi neuromorphic chips (the name of an underwater volcano near Hawaii Island) . The company will use Pohoiki Springs for machine learning applications, including a possible overview of the spread of the coronavirus.
The launch of Pohoiki Springs comes a little later than expected . Intel’s roadmap , revised last July, expected it to be delivered to researchers by the end of the year. Mike Davies, director of the Neuromorphic Computing Lab at Intel, described Pohoiki Springs as having the computer intelligence of a mole rat, with the equivalent of 100 million artificial neurons (80 to 100 billion for a human) . Available in cloud mode, the Pohoiki Springs system “will be made available to members of the Intel neuromorphic research community (INRC), extending their neuromorphic work to solve larger and more complex problems”,
Intel thinks of Loihi in the same way that it treats Core CPUs: as a fundamental architecture that can be increased and reduced according to the needs of the application. For example, a Loihi two-channel system called Kapoho Bay is used for edge processing, directly processing data from cameras and other sensors and interpreting this signal. A Nahuku platform combines 32 Loihi chips together.
“So Loihi is sort of reaching an average equilibrium point, where we have 128 cores, we can adapt with several chips to go more upscale,” said Mr. Davies during a conference call with analysts and journalists. “But it’s also not too big to be useful as an edge device. ”
The purpose of the Loihi chip is to be used as a research tool for machine learning. It simulates the functioning of the human brain, in order to determine whether this approach is in fact a more efficient way of performing ML functions. such as identifying and categorizing images based on their content. Earlier this week, Intel taught Loihi how to feel .
According to Davies, another function for which Loihi could be used would be to model what are called “small world” statistical models. Small world models are currently interesting because they model social networks in the real world, where people interact and re-interact with others.
These models “could very well be used to model different scenarios of how coronaviruses could spread around the world depending on how you cut or slow social connections or interactions,” said Davies … a real-world study of how “social distancing” actually works to slow the spread of disease.
Such studies could be done by an x86 chip like Intel’s Xeon. But Loihi is designed to be much more efficient in these types of scenarios, consuming only 100 watts in a server form factor of 5U. “So we’re very excited by these results that we’ve seen so far,” said Mr. Davies. “And we want to see what we can achieve by exploring this direction on a scale.” Ultimately, said the manager, Loihi could end up in a laptop, where Loihi kernels or hearts could be used to process a user’s visual cues and “allow more natural interaction with your computer,” said Mr Davies.