AKIDA benchmarking?, page-127

I agree with what you're saying - in its infancy there isn't an agreed standard for comparing performance but something is better than nothing. And tests against hardware are not the same as development environments. Latency/Power/Accuracy(and cost) will determine the advantages of Akida over its competitors which is why enthusiasts (except the weirdos who inhabit this forum) are eager for some latency/power comparisons to eventuate.

Here's an example of Intels benchmarks in which power/latency are the key metrics against datasets which are not industry standard (as industry standard do not exist)

https://www.eetimes.com/intel-benchmarks-neuromorphic-chip-against-ai-accelerators/

Brainchip developers will ABSOLUTELY be trying to benchmark against Intel open sourced tests [referenced in link above])

I reeeeally want to know what their findings are.

Despite his huffing and puffing,@BarrelSitter was a good sport and provided what seems to be Brainchips only published benchmark citing latency. All of which are a tidy 10 FPS and 100ms across disparate tests in ADE.

Interesting that Akida felt the need to include speed and cited FPS. Some posters here said speed and FPS doesnt matter (and they invested in company designing computer chips LOL, god help us....)