BRN Discussion Ongoing

Diogenese

Top 20
Hi Rach,

This is an endorsement of RAG (Retrieval Augmented Generation), or rather, RAG can support this proposal.

You don't need all the information in the world to navigate through life. Past experience can provide a guide for new experiences.
 
  • Like
  • Fire
  • Love
Reactions: 7 users
Can you please all stop this Trump-discussion?! This forum is about BrainChip! Find your own forum for

What a load of rubbish. Look at you calling Trump a pedo with no proof whatsoever.. your mate Joe Biden was letting the kiddy fiddlers in with the rapists and murders and god knows what else.. that’s the thing with you lefties you have no common sense
Ummm, i didn’t call Trump a pedo, you just did. Now c’mon.. really, wouldn’t common sense tell you he is at least protecting them and involved. Thats it from me, sorry all, back to BRN.
 

Frangipani

Top 20

View attachment 92050

G'day sleepy,

Back in April, Rinaldi was a top 50 holder (50th spot), with just under 2.2 million shares. Still some skin in the game there. I think his comments will ring true for many of us.

BRN Top 50

Don't get me started on the CR that came from nowhere......

Good news is the SP is showing signs of life so hopefully signed deals are imminent. (As in next week)


As this year’s 50th spot in the Top 50 holder list (28 April 2025) is a shareholder with just over 2.3 million shares, we don’t know how much skin in the game our former NED Neil Rinaldi still has - he could theoretically have held on to all of those shares he still had last April, see @FJ-215’ s November 2024 post 👆🏻.

What we do know, though, is that he is not at all happy with Sean Hehir and the Board, given that he felt the need to vent publicly on LinkedIn:



A65F6113-D84D-4E79-912D-E88D0E6D6729.jpeg



E18CCB28-5D39-4F11-8838-C04E2A26AE56.jpeg
 
  • Like
  • Wow
  • Fire
Reactions: 17 users

Guzzi62

Regular

As this year’s 50th spot in the Top 50 holder list (28 April 2025) is a shareholder with just over 2.3 million shares, we don’t know how much skin in the game our former NED Neil Rinaldi still has - he could theoretically have held on to all of those shares he still had last April, see @FJ-215’ s November 2024 post 👆🏻.

What we do know, though, is that he is not at all happy with Sean Hehir and the Board, given that he felt the need to vent publicly on LinkedIn:



View attachment 92087


View attachment 92089
LOL-LOL.

I couldn't care less what he says and thinks.

He is properly behind some of the pathetic downrampers over at the crapper.

Let Sean stick to his 5 years plan & 9 mill income this year. If not, his next AGM will be very hard and maybe the last one for him.

Until then, I keep quiet about any criticism and let the man do his work.
 
  • Like
  • Love
  • Fire
Reactions: 15 users

Frangipani

Top 20
Hey, we small fry will get a Social Media Marketing Intern, though… 😉


View attachment 86378

View attachment 86379


Hopefully they won’t accidentally hire the same person who claimed the following on Reddit the other day, without providing any sources whatsoever.

Most likely a BRN shareholder who wanted to spruik Akida and drive up the share price with some fabricated “evidence”.

Unsubstantiated hype like that is NOT helpful…




View attachment 86385
View attachment 86386 View attachment 86387


In case you missed the news:

“On June 1, Russia’s Military Transport Aviation Day, a significant holiday for the Russian armed forces, the Security Service of Ukraine (SSU) carried out a bold and unprecedented coordinated drone strike deep inside Russian territory. The operation targeted four strategic air bases and delivered a major blow to Moscow’s long-range bomber fleet. Codenamed “Spider’s Web”—or simply “Web”—the operation was named for its wide geographic coverage across remote Russian locations previously thought to be beyond the reach of Ukraine’s long-range drone capabilities.”



Meet our Social Media Marketing Analytics Intern Kylie Situmorang, who started her internship with BrainChip last month:


9FEAE597-2644-4E2E-A53A-F85583A75375.jpeg
87E6E7DB-CC9F-4E34-830C-6AF03553D665.jpeg
81B35767-0DE4-4F15-BD9C-1F4A7E563867.jpeg
 
  • Like
  • Thinking
  • Fire
Reactions: 9 users

CHIPS

Regular
  • Like
Reactions: 3 users

BrainShit

Regular
From the 30min on time frame Dr Harbor talks about the start up of Neuromorphic computing and he goes on to say
"At the end of the day after the long haul we are going to see the analog neuromorphic like processor is going to win out."
We are digital but we use Spikes instead of 1's and 0's is my understanding.
Now I'm confused. Any further explanation would be gratefully accepted.

Yes, Akida uses spikes and is fully digital. The signal processing can (depending on the used processor) vary between analog, digital or mixed.
E.g. in convolutional neural networks (CNNs) and spiking neural networks (SNNs) each network has specific advantages and disadvantages:

Analog signal processing

Advantages:
- Provides natural and continuous signal transission, which is often perceived as warmer.
- Easier implementation in certain applications.
- Can often be operated with less energy because they process continuous signals and do not require digital conversions.
- Enable a faster response to changes in the signal, which is advantageous for applications with real-time requirements.

Disadvantages:
- Susceptible to interference and noise, which can affect signal quality.
- Lower confidentiality and security compared to digital signals.
- Limited flexibility: Analog signals are often less flexible to customize and process than digital signals, which can be easily edited and stored.
- Scaling difficulties: When processing large amounts of data, analog signal transmission can be less efficient because it does not offer the same data compression and analysis capabilities as digital systems.

Digital signal processing

Advantages:
- Higher accuracy and better anti-interference capability, resulting in higher signal quality.
- Flexibility in processing and integration into digital networks.

Disadvantages:
- Requires more bandwidth and more complex technical requirements, especially for synchronization.
- Can sound unnatural when quality is poor and is more prone to latency issues.

Mixed-signal processing
Mixed-signal systems combine analog and digital processing, which allows for a balance between the advantages of both technologies, but also increases complexity. >> Our competitor Innatera offers this with TSP1 & Pulsar

... which signal processing you need/want depends of the used network and task / product requirements and you budget.
 
Last edited:
  • Like
  • Fire
Reactions: 11 users

Tothemoon24

Top 20
Apologies if this has been shared, podcast with Sean from September




 
  • Like
  • Fire
  • Love
Reactions: 8 users
LOL-LOL.

I couldn't care less what he says and thinks.

He is properly behind some of the pathetic downrampers over at the crapper.

Let Sean stick to his 5 years plan & 9 mill income this year. If not, his next AGM will be very hard and maybe the last one for him.

Until then, I keep quiet about any criticism and let the man do his work.
Must be your boss then 😂
 
  • Haha
Reactions: 1 users
She started in September? I haven't noticed any difference yet.
No matter who we take on we never see any difference apart from Alf 👍
 
  • Like
Reactions: 4 users

Frangipani

Top 20
New LinkedIn post on keyword spotting.
I think the drone video is the one, though, which is a few years old?




426C32A1-85BA-48AB-B178-B73C8F28EFD1.jpeg





4DF6CAF1-B4B1-45A6-AFCD-CB9FA0A77102.jpeg

2CECBE3F-E005-47D4-A874-8AF83AE12578.jpeg

6B8FC64B-D1AD-495C-BC6C-E2B8B8A86BB0.jpeg

4FDB400E-E841-4CEE-A6B0-22D12B786A2A.jpeg

DDAB14A9-BCD8-457B-BE9E-87591FA9267C.jpeg

0AE53193-EE01-410A-BBE8-19B35957696B.jpeg
 
Last edited:
  • Like
  • Fire
  • Love
Reactions: 19 users

Frangipani

Top 20
Daniel Andersson, Project Manager at Frontgrade Gaisler, will be the one presenting “GRAIN – towards event-based AI in space” at EDHPC 2025 tomorrow:


View attachment 91987


C582E8A3-288D-4165-BC0D-42C327843ADC.jpeg





Laurent Hili from ESA and Juan P. Cobos Carrascosa from Frontgrade Gaisler co-chaired the session AI Engines & Neuromorphic

7E015A68-BF1D-4B2E-BD47-53FEB54ADD6D.jpeg




7EEF21FE-240D-44D7-A50E-A8B96503FE73.jpeg

[…]

A518C4A8-FBAC-4EFB-8F83-3950CE661D3A.jpeg

7CF76B92-A6AD-489A-8B1B-A7CC9A156FD6.jpeg
 
Last edited:
  • Like
  • Love
  • Fire
Reactions: 12 users

IloveLamp

Top 20
If you don't like chat gpt look away now


1000012545.jpg
 
  • Love
  • Like
  • Fire
Reactions: 8 users

7für7

Top 20
LinkedIn post


 
  • Fire
  • Love
Reactions: 2 users

Terroni2105

Founding Member
If you don't like chat gpt look away now


View attachment 92116


I just asked it who DJI is and it said:

DJI, which stands for Da Jiang Innovations, is a Chinese technology company that has become the world's largest manufacturer of commercial and consumer drones. Founded in 2006 by Frank Wang, a Chinese aerospace engineer, it is based in Shenzhen, China.

DJI controls over 70% of the global drone market. The US government has raised national security concerns over potential data leaks from DJI drones, and in 2021 placed the company on an investment blacklist. DJI has faced lawsuits, including a successful one by Textron over a patent infringement for its automatic hovering capabilities.

😳🤔

Having said that, I remember seeing this video years ago so since then hopefully we’ve moved on to a different drone company (like the military etc)
 
  • Thinking
  • Like
  • Fire
Reactions: 6 users

7für7

Top 20
The more news appears, the more aggressively some market participants short and manipulate. Out of fear that the price could rise on positive developments, it no longer seems to be primarily about profit for them, but rather personal motives — in the 20–22 cent range hardly anyone makes a meaningful profit. It puzzles me that this isn’t being investigated and potential manipulation isn’t being held accountable.

BRING IT ON BRAINCHIP!!!!! LET THEM BURN!!!!

Ashton Kutcher Burn GIF
 
  • Like
Reactions: 3 users
Just looked at the buyers and sellers lines and I wonder if there are any retail orders in it 😂
 
  • Like
  • Sad
Reactions: 3 users

DK6161

Regular
Just looked at the buyers and sellers lines and I wonder if there are any retail orders in it 😂
My buy order of 1000 units got wiped so quickly. Should've waited
 
  • Haha
Reactions: 1 users
It would appear the gentlemen below from VW running more projects with Akida. Uploaded to GitHub yesterday.

Results look strong.

Fernando Sevilla Martínez​





SevillaFe/EcoEdgeAI-akida-macPublic

A comprehensive workflow for comparing energy efficiency between conventional hardware (Mac M-series GPU/CPU) and neuromorphic hardware (Akida on Raspberry Pi 5) for autonomous driving steering angle prediction.

SevillaFe/EcoEdgeAI-akida-mac​




Name
SevillaFe SevillaFe
yesterday

LICENSEyesterday
README.mdyesterday
requirements_mac.txtyesterday
requirements_rpi5.txtyesterday
workflow_guide.mdyesterday

Repository files navigation​

EcoEdgeAI-akida-mac​

A comprehensive workflow for comparing energy efficiency between conventional hardware (Mac M-series GPU/CPU) and neuromorphic hardware (Akida on Raspberry Pi 5) for autonomous driving steering angle prediction.
License: MIT Python 3.8+ TensorFlow 2.16+

🎯 Project Overview​

This project provides a complete pipeline to:
  • Train deep learning models (PilotNet, LaksNet, MiniNet) for steering angle prediction
  • Benchmark inference performance on conventional hardware with CodeCarbon energy tracking
  • Convert models to neuromorphic format using Akida
  • Benchmark neuromorphic inference with TC66 USB power meter
  • Generate comprehensive eco-efficiency comparisons

Key Research Questions​

  • How much energy does neuromorphic computing save?
  • What is the accuracy trade-off?
  • What is the latency difference?
  • Which architecture is most efficient for edge deployment?

📊 Results Preview

Our experiments show:
  • Energy Efficiency: Up to 76% reduction in energy consumption per inference
  • Latency: 40-50% faster inference on neuromorphic hardware
  • Accuracy: Minimal degradation (<10% MSE increase)
  • Overall Efficiency: 280%+ improvement in accuracy-per-energy ratio

🛠️ Hardware Requirements​

Mac (Training & Benchmarking)​

  • MacBook with M-series processor (M1/M2/M3)
  • 16GB+ RAM recommended
  • macOS 12.0+

Raspberry Pi 5 (Neuromorphic Benchmarking)​

  • Raspberry Pi 5 (4GB/8GB)
  • BrainChip Akida neuromorphic processor board
  • TC66/TC66C USB power meter
  • 32GB+ microSD card
  • Active cooling recommended
 
  • Fire
  • Like
  • Love
Reactions: 7 users

7für7

Top 20
The pump Sold again within two days… and the moment it starts rising, you immediately get a ticket. Just ridiculous what’s going on.
No normal positive news will bring us back to even 25-30 cent… without revenue we will be the play ball for shelters and day trader… very simple
 
  • Sad
Reactions: 1 users
Top Bottom