2024 BrainChip Discussion, page-3389

At least five months ago it was publicly known that Brainchip AKIDA 2.0 had these capabilities. Since that time there have been a number of other references yet this poster chooses to promote an article where even the author a techie from Tom’s Hardware states:

“Though we are told that the KAISTC-Transformer chip can do the same LLM processing tasks as one of Nvidia'sbeefy A100 GPUs,none of the press nor conference materials we have provided any directcomparative performance

metrics. That's a significant statistic,conspicuous by its absence, and the cynical would probably surmise that aperformance comparison doesn't do the C-Transformer any favours....

You can see that the C-Transformer is currentlyfabbed on Samsung's 28nm process and has a die area of 20.25mm2. It runs at amaximum frequency of 200 MHz, consuming under 500mW. At best, it can achieve3.41 TOPS.

At face value, that's 183x slower than theclaimed 624 TOPS of the Nvidia A100 PCIe card (but the KAIST chip is claimed touse 625x less power). However, we'd prefer some kind of benchmark performancecomparison rather

than look at each platform's claimed TOPS.’

Jean-Luc ChatelainJean-Luc Chatelain• 3rd+• 3rd+Founder & Managing Director @ Verax Capital Advisors | AI TransformationFounder & Managing Director @ Verax Capital Advisors | AITransformation

5mo • Edited • 5mo • Edited •

Follow

As my followers (humble thankyou for this) know, I am a Neuromorphic computing fan. The team at BrainChip is innovating again in a big way to bring GenAI & LLM to the edge with their 2nd generation platform.

https://lnkd.in/ecRNmMMk

#generativeai #genai #vision #edgematters #ai #artificialintelligence #neuromorphic #edgecomputing

BrainChip Makes Second-Generation Akida Platform Available to AdvanceState of Edge AI Solutions

brainchip.com • 4 min read

https://www.linkedin.com/posts/jlchatelain_brainchip-makes-second-generation-akida-platform-activity-7118510710438944768-xvAr

My opinion only DYOR

Fact Finder