BRN Discussion Ongoing

D

Deleted member 118

Guest
 
  • Like
  • Wow
Reactions: 8 users

Slymeat

Move on, nothing to see.
so that no one forgets how great Brainchip is, here it is again

Nice post from the class forum. Translation from memory!


Webinar: Brainchip is specifically captured only with the current de facto standard NVIDIA Jetson.

The two techniques were tested with the same Nviso software on identical systems with the latest software/drivers. Both systems were tested on 5 standard models currently used by Nviso (gestures, etc.). It is interesting that the main customers who use these models include Siemens (in the medical field), Panasonic (private customers) and ZF (mobility).

The results can be read on the charts. Summarized:

- Akida runs at a lower clock compared to NVIDIA (300 MHZ compared to NVIDIA with approx. 900 CPU and approx. 1400? GPU, which leads to a significantly lower power consumption on the part of Akida
- In terms of output, Akida achieves an average of over 1,000 FPS for all models - well ahead of NVIDIA, which neither the GPU nor the CPU can achieve in the best model. (This is the performance index)
- Response time: Akida is below 1 ms for the models. Again, NVIDIA doesn't match this on the GPU or CPU side.
- Memory side - and this is where the money comes into play as memory is expensive - Akida only needs 1/5 of NVIDIA. Akida is under 1 MB using all 5 models cumulatively

Finally: Akida has at least the same performance (the emphasis is on at least) with significantly lower acquisition costs and running costs (energy consumption). The low energy consumption alone at this level of performance takes Akida into other spheres - apart from one-shot learning etc.
My take away from this video was that the models were written by Nviso to run on the NVIDIA GPU and the CPU, and were probably optimised for that hardware. It is, after all, what Nvisio have been doing for a decade. Then the models were quickly ported (Tim’s words, and I think he said it took only a matter of hours) to Akida where they blew away both the GPU and CPU in all aspects. I was extremely impressed. I don’t consider the performance as similar, Akida was a clear winner in 4 of the 5 models with only facial gesture monitoring being comparable.

Tim was overly obvious in his choice of words in talking generically about neuromorphic computing. But it was also so blatantly obvious that every time he used the generic term, he may as well have said Akida. Akida, after all IS the ONLY SNN hardware out there and is the ONLY hardware he did his testing on. It was with Akida hardware and software that his software successfully integrated, and not with neuromorphic computing, as was stated. Although the former does imply the later.

The first half of the video is basically a testament to how good Akida is. And the second half was a look into all the places Akida can be used.

I assume a huge part of this video was to introduce their SDK. And well it should be, that’s how,they are going to,get developers interested and get people to play with Akida. Unfortunately, I found that part of the video quite boring. I believe it wasn’t clearly thought through. That sort of stuff should never be done live, it should be pre-recorded and presented as smoothly as possible. Then you can enhance the video with pop-ups explaining what is going on. Watching textnon a screen never makes for great viewing.
 
  • Like
  • Love
  • Fire
Reactions: 35 users
D

Deleted member 118

Guest
  • Like
  • Fire
Reactions: 2 users

TopCat

Regular

AI-Driven Decision-Making​


Palantir Edge AI is Palantir’s AI orchestration and sensor fusion engine that runs on disconnected, remote endpoints. It enables autonomous decision-making for on-hardware models consuming real-time sensor, radio, acoustic, geo-registration, and time series data. Extremely lightweight and power efficient, Palantir Edge AI brings AI to drones, vehicles, buildings, oil rigs, aircraft, ships, wind turbines, robots, satellites — and more.





Now this would be a great rumour to start! Just need to find that solid link to Palantir 🤔
 
  • Like
  • Thinking
  • Fire
Reactions: 10 users

MrNick

Regular
 
  • Like
  • Fire
  • Love
Reactions: 16 users
I buckled yesterday… sold what very little I had left in LKE at a small loss.

Bought extra 900ish BRN. I know that’s peanuts to some holdings here of course.

Tax return should clear today/Monday (hurry up!!! The bus is starting to roll…) and hope to make it to my (very modest) target of 10k shares before long. Then my Q2 bonus will come in end of August. All this is usually my play money. So no savings, home budgets or contingencies where or are being hurt in the process 🤣😂

Even considering to sell to some guitars that have sat in their cases for too long. Nows the time… all aboard!!
 
  • Like
  • Love
  • Fire
Reactions: 58 users
D

Deleted member 118

Guest
did you know that ? always thought peter holds the patents! according to the link he ceded them to brn 🙄
I wondered what you was pointing out now I know 👍
 
  • Like
  • Haha
Reactions: 5 users

Slymeat

Move on, nothing to see.
  • Like
  • Haha
Reactions: 5 users

Gemmax

Regular
Getting their arse handed to them atm. Bloody tipped them to
I'm Parra supporter from way back. Pretenders not contenders I suspect.
 
  • Sad
  • Like
Reactions: 2 users
Wake Up Coffee GIF by good-morning

Good morning my fellow and fellowets chipsters:ROFLMAO:
 
  • Haha
  • Like
  • Love
Reactions: 17 users
This seems to indicate that PVDM signed over his patents to BrainChip back in 2015?
that gives the basis for speculation again 😄😄😄😄
 
  • Like
Reactions: 1 users

stuart888

Regular
Think this image shows a nice easy visual representation.

Note the first short sales spikes and what happened to the SP straight after then look at the latest last spike and what the SP is doing now....burn ya bastards burn :cautious:


Short Sales (Official ASX Data)​

This table shows the daily short sale activity as reported by the Australian Stock Exchange (ASX)

Update Frequency: Daily


View attachment 12155

View attachment 12156
Thanks to @Fullmoonfever and to all the folks that highlight the shorting. While I focus on the technology, spike adoption, and the business strategy for the long-term; it is nice to be educated on all aspects. It all adds up to more knowledge, so thanks to all the short message people. Great charts, lots of effort.

For Fullmoon, Tom Petty played at the Don Cesar Beach Resort in 1985. I am staying there this weekend, as it is near me. This old video might be interesting for you, given your icon image!


 
  • Like
  • Fire
Reactions: 16 users
F

Filobeddo

Guest
  • Haha
Reactions: 2 users
What, no hair of the dog? 😉
well Kinda woke up with my dog sleeping on my chest.
still a bit early for me but is this the patent in above document
 
  • Like
  • Haha
Reactions: 5 users
Anil liked post on linkedin regarding this.
":

Why Deeply Quantized Neural Networks Matter?​

The industry’s challenge was to find a way to simplify neural networks to run inference operations on microcontrollers without sacrificing accuracy to the point of making the network useless. To solve this, researchers from ST and the University of Salerno in Italy worked on deeply quantized neural networks. DQNNs only use small weights (from 1 bit to 8 bits) and can contain hybrid structures with only some binarized layers while others use a higher bit-width floating-point quantizer"
 
  • Like
  • Fire
Reactions: 9 users
  • Like
  • Love
Reactions: 12 users
F

Filobeddo

Guest
well Kinda woke up with my dog sleeping on my chest.
still a bit early for me but is this the patent in above document

How many did he have?? ;)

1658440110258.png


You do know that one dog beer is like 15 human beers ;);) 😂😂😂😂
 
Last edited by a moderator:
  • Haha
  • Like
Reactions: 11 users
Top Bottom