Share
1,441 Posts.
lightbulb Created with Sketch. 878
clock Created with Sketch.
02/03/21
23:44
Share
Originally posted by CHAPPY99:
↑
rayz. I believe I got you right !!! End date of Vorago CNN RNN PROCESSOR Phase 1 Hardening on 1st march 2021 !!!!! .........................................................................................................................................................................................................................The ultimate goal of this project is to create a radiation-hardened Neural Network suitable for Ede use. Neural Networks operating at the Edge will need to perform Continuous Learning and Few-shot/One-shot Learning with very low energy requirements, as will NN operation. Spiking Neural Networks (SNNs) provide the architectural framework to enable Edge operation and Continuous Learning. SNNs are event-driven and represent events as a spike or a train of spikes. Because of the sparsity of their data representation, the amount of processing Neural Networks need to do for the same stimulus can be significantly less than conventional Convolutional Neural Networks (CNNs), much like a human brain. To function in Space and in other extreme Edge environments, Neural Networks, including SNNs, must be made rad-hard.Brainchiprsquo;s Akida Event Domain Neural Processor (www.brainchipinc.com) offers native support for SNNs. Brainchip has been able to drive power consumption down to about 3 pJ per synaptic operation in their 28nm Si implementation. The Akida Development Environment (ADE) uses industry-standard development tools Tensorflow and Keras to allow easy simulation of its IP.Phase I is the first step towards creating radiation-hardened Edge AI capability. We plan to use the Akida Neural Processor architecture and, in Phase I, will: Understand the operation of Brainchiprsquo;s IP Understand 28nm instantiation of that IP (Akida) Evaluate radiation vulnerability of different parts of the IP through the Akida Development Environment Define architecture of target IC Define how HARDSILreg; will be used to harden each chosen IP block Choose a target CMOS node (likely 28nm) and create a plan to design and fabricate the IC in that node, including defining the HARDSILreg; process modules for this baseline process Define the radiation testing plan to establish the radiation robustness of the IC Successfully accomplishing these objectives:Establishes the feasibility of creating a useful, radiation-hardened product IC with embedded NPU and already-existing supporting software ecosystem to allow rapid adoption and productive use within NASA and the Space community.\n\n\n\n\t Creates the basis for an executable Phase II proposal and path towards fabrication of the processor.\n\n ............................................................................................................................................................................................................................. .
Expand
Hey Chappy (and others), going by the line in here with 'akida neural processor architecture ", is it reasonable to read into that that NASA's end goal is going to be an IP license rather than chips? I get they're using our chips to start and test on, but now I think about it I guess it would make sense for most of their chips to be custom designed - they have the budget, expertise, and space/weight is critical so having everything customised... They also talk about 'understand 28nm instantiation of IP ' (akida chip) followed by 'Define architecture of target IC' and '...each chosen IP block' I suppose a follow up question that raises for me is who are we expecting to be using an actual Akida chip rather than just the IP? (either directly or via a chip from Renesas or similar)