2020 BRN Discussion, page-21398

Hi BarrelSitter
When this information regarding Intel and Tensor was first posted I posted that they were two entirely different things but no one wanted to listen. I was cheating a little bit because I had previously done a deep dive into Loihi and its variants. The Intel NCS2 is not a processor it is an accelerator which allows one to build software and algorithms for implementing neural computing functions into your system. I suggested in my post that this was the case and that perhaps it was that Tensor intended using the Intel developer environment to build the software/algorithms necessary to support its implementation of AKD1000 into the unique CubSat roles it proposed to NASA. It has been commented upon numerous times that Tensor Partners do not have massive resources if any in the neuromorphic system development area and that Brainchip has a limited number of engineers available to support their EAP partners. In saying this I do not diminish the expertise of the Tensor Partners because they are in fact both rocket scientists just not neuromorphic specialists.

This time I have extracted the following from Intel web sites to support my contention and in particular I draw your specific attention and that of uiux and others with tech knowledge to the MINIMUM system requirements which you need to support the use of the Intel NCS2 accelerator which given this is being developed for CubSat use will require a Saturn Rocket to get it into orbit so clearly Tensor has in mind some other neuromorphic/processor system on a chip to implement the developed software with when incorporating it into CubSat.

https://www.intel.com/content/dam/support/us/en/documents/boardsandkits/neural-compute-sticks/NCS2_Product-Brief-English.pdf

• 
  USB plug & play AI device for deep learning inference at the edge.
• Included ItemsThe Intel® NCS2 is built on the Intel® Movidius™ Myriad™ X VPU featuring 16 programmable shave cores and a dedicated neural compute engine for hardware acceleration of deep neural network inferences.

Minimum System Requirements

CPU:

• 6th to 9th generation Intel® Core™ and Intel® Xeon® processors
• Pentium® processor N4200/5, N3350/5, or N3450/5 with Intel® HD Graphics

Graphics:

• 6th to 9th generation Intel Core processor with Iris® Pro graphics and Intel HD Graphics
• 6th to 9th generation Intel Xeon processor with Iris Pro graphics and Intel HD Graphics (excluding the e5 product family, which does not have graphics)

3 December, 2020

During Intel's virtual Lab Day, on Thursday, the tech giantrevealed Loihi chip improvements in voice command recognition, gesturerecognition in artificial intelligence (AI) applications, image retrieval,search functions, and robotics.

Neuromorphic computing aims to use computer science to propelrule and classical logic-based AI builds into more flexible systems thatemulate human cognition, including contextual interpretation, sensoryapplications, and autonomous adaptation.

Intel says that neuromorphic computing focuses on emulating thehuman brain and implementing stable probabilistic computing, which creates"algorithmic approaches to dealing with the uncertainty, ambiguity, andcontradiction in the natural world" -- just like humans are capable of.

However, speaking to attendees of the virtual event, Rich Uhlig,VP and Director of Intel Labs added a caveat: progress in neuromorphiccomputing has "come at the cost of ever-increasing power consumption,[...] posing challenges for AI and the democratization of AI" -- abottleneck hurdle that the company wants to overcome.

The Loihi research chip, containing 128 cores and fabricatedthrough 14nm process technology, is Intel's contribution to the field. Makingits original debut in 2017, the processor contains 130,000 'neurons' suitablefor use in spiking neural networks (SNNs) that send pulsed signals --containing encoded information -- which 'spike' and stimulate other neurons ina neural network.

The expense of neuromorphic chip design and production willlikely keep neuromorphic computing applications at the edge or fornon-cost-sensitive systems, Intel says, but the firm expects innovation in thefield to eventually drive down the price of adopting neuromorphic technologiesand push it firmly into the commercial sector.

In the end, Intel wants to create an architecture focused onpower efficiency and speed, with little to no detriment to performance, aconcept the company calls an "order of magnitude gains."

https://www.zdnet.com/article/intel-releases-new-benchmarks-progress-on-loihi-neuromorphic-computing-chip/

My opinion only DYOR.