Renesas, page-54

Thanks @Fact Finder and @BarrelSitter actually what I am after is a bit more simple...the background is that I have been in correspondence with the Principal director from one of the larger consulting firms that completed a review (a while back) on neuromorphic advancements and how vital this technology will be...I asked why they had not included BrainChip as they currently have a product offering, but instead focused on Intel and IBM. The answer I got back was promising...

"We do know about Brain Chip and are interested in learning more. My Neuromorphic Computing team is small; we cannot explore everything at once. The article was not meant to be a comprehensive landscape analysis, but rather a first article, focused more on conceptual introduction with the vendors reflecting what we have so far had time to get smart on. Brain Chip is on our radar, and may very well get more of our hands-on attention in the near future."

I was hoping to provide some high level details to spark increased interest e.g.

• Neurons: Akida 1200000, TrueNorth 1000000, Loihi 131072,
• Fully compatible with popular ML technologies like Tensorflow
• Ultra low power consumption and heat dissipation, less than 200 microwatts.
• Optional event-based convolution and pooling assets in each reconfigurable layer, enabling execution of event-based CNNs as well as SNNs on the same hardware, simultaneously.
• Possible to implement a large range of different networks on the same hardware.
• On Chip Learning, Possible to perform incremental learning after training

Basically, things that we think setAKIDA apart from the rest of the industry, in the hope that this may encouragethem to do some further validation. It will be their job to validate orbenchmark all findings.

cheers