Intel's Neuromorphic Chip Scales Up (and It Smells)
Neuromorphic chips attempt to directly mimic the behavior of the human brain. Intel, which introduced its Loihi neuromorphic chip in 2017, has just announced that Loihi has been scaled up into a system that simulates over 100 million neurons. Furthermore, it announced that the chip smells. (That is to say: it's able to smell. To a nose, it probably just smells like a computer chip.)
Loihi is Intel's fifth-generation neuromorphic chip. It packs 128 cores – each of which has a built-in learning module – and a total of around 131,000 computational "neurons" that communicate with one another, allowing the chip to understand stimuli. The new system, Pohoiki Springs, contains over 100 million of those computational neurons. It consists of 768 Loihi chips, mounted on Intel Nahuku boards in a chassis that Intel describes as "the size of five standard servers," and a row of Arria10 FPGA boards. By contrast, Kapoho Bay, Intel's smallest neuromorphic device, consists of just two Loihi chips with 262,000 neurons.
"Pohoiki Springs scales up our Loihi neuromorphic research chip by more than 750 times, while operating at a power level of under 500 watts," said Mike Davies, director of Intel's Neuromorphic Computing Lab. "The system enables our research partners to explore ways to accelerate workloads that run slowly today on conventional architectures, including high-performance computing systems."
Also at The Next Platform, EE Times, and Wccftech.
(Score: 2, Interesting) by Anonymous Coward on Wednesday March 25 2020, @03:40PM (1 child)
So what is it?
"neurons" can be anything from an FPGA LUT to an array of 16 bytes in a ram cell to a monstrous construction several megabyte in size, depending on model used and level of detail, as well as what subset of known biological features one chooses to implement.
link to whitepaper: https://www.researchgate.net/publication/322548911_Loihi_A_Neuromorphic_Manycore_Processor_with_On-Chip_Learning [researchgate.net]
"oihi is a 60 mm 2 chip fabricated in Intel’s 14nm process that advances the state-of-the-art modeling of spiking neural networks in
silicon. It integrates a wide range of novel features for the field, such as hierarchical connectivity, dendritic compartments, synaptic delays, and
most importantly programmable synaptic learning rules"
in the report:
"We adopt a variation of the well-known CUBA leaky-
integrate-and-fire model that has two internal state variables, the
synaptic response current u i (t) and the membrane potential v i (t) ."
So they model a neuron WITH 2 VARIABLES.
Sure, it can be used for fancy pattern recornition. But it aint a breakthrough, or "an attempt to directly mimic the behavior of the human brain".
To me, it resembles "an attempt to directly mimic IBM and their equally useless spiNNaker arch"
https://en.wikipedia.org/wiki/SpiNNaker [wikipedia.org]
(Score: 0, Spam) by tanusaha on Thursday March 26 2020, @12:25PM
I am very impressed with your article. I love the way you write this article. Thank you for sharing this wonderful post with us. Also, read this: https://www.techpout.com/best-duplicate-photo-finder-and-cleaner-software/ [techpout.com]