ELI5, Bananas, and what Project Phasor is NOT…
Someone asked for an ELI5 on Project Phasor in the Open Neuromorphic discord channel. Copying my reply here:
“““
It intends to be like the trunk of a banana plant, to fruit a bunch of bananas (i.e. the proposed 5 areas of focus) all at once and then wither away. Notably, the fruit will be of the rare Gros Michael variety (large scale neuromorphic training) rather than the common Cavendish variety (neuromorphic inference at the edge). The Panama disease (short term vision and diffuse efforts) incentivizes industry to produce Cavendish even though Gros Michael is what everyone actually wants to work on. Project Phasor will also eradicate Panama disease by focusing on production-quality software (compilers and heterogeneous integration), removing artificial hardware constraints as much as possible (by encouraging a zoo of neuromorphic VMs), and encourage a zoo of higher level frameworks (by explicitly NOT trying to force a neuromorphic framework standard, which several have tried and failed to do since it is still to early). The “roots” of the banana plant will be the joint research fund, which will be used to produce future plants of the Gros Michael variety. By throwing everything at the wall at the same time, Project Phasor hopes to use the power of coincidence to trigger positive feedback that cascades into a stable attractor basin of neuromorphic awesomeness.
”””
How things actually shape out will depend on the community. We have a lot of meetings slated for January, to be capped off with an Open Neuromorphic hosted presentation. Register your interest here to prioritize the NeuroAI and neuromorphic issues best solved together!
Link in reply.
#projectphasor #ai #ml #funding #compilers #virtualmachines #heterogeneouscomputing #datacenterscale #neuromorphic #neuroai