BRN Discussion Ongoing

alwaysgreen

Top 20
I respectfully disagree, in that running changes, happen with auto makers all the time, since year dot.
The only thing that is "fixed" is the main body styling.

We have no idea really, how long they've been working on it, with MegaChip’s.
It was possibly before, MegaChips even signed the IP deal (we know that Mercedes would have been working with BRN, well before).
They wouldn't have done so, if there wasn't already significant traction, to justify the 2 million dollar licence fee (and in fact, I think MegaChips, stated something to this effect, in the original agreement, as to "customers" that Brainchip, already had)..

It's all speculation, but shooting the idea down as a matter of fact, is unreasonable, in this case, in my opinion 😛
Absolutely agree that we are just speculating. Not trying to admonish your opinion and I am not saying I am definitely correct (I clearly said probably and in my opinion a few times 🤪). I sincerely hope I am wrong and it is being implemented but I don't believe it is. I think we all get caught up in the hype a little sometimes. We all want BRN to build our wealth, and the faster the better haha
 
  • Like
  • Love
Reactions: 11 users

Learning

Learning to the Top 🕵‍♂️
Implementing a new system in a car takes more than a few months though. The features of a new model are locked in well in advance of production. I can't see how it would be logistically possible to roll out Akida into a model that has just hit the production line given they have only implemented it previously in one concept car (in January). I like to dream too but I think you probably answered your own question. Megachips probably haven't even finalised the chips. Don't get me wrong, I would love for every model to have this implemented moving forward but concept car features trickle down to production cars over year, not months.
Just a question to the more informed.

Is its possible to implement Akida software first?
We know the Akida hardware will take time to implement, but the software is simpler to implement into current hardware.

Learning
Its great to be a shareholder.
 
  • Like
  • Fire
Reactions: 5 users
Just a question to the more informed.

Is its possible to implement Akida software first?
We know the Akida hardware will take time to implement, but the software is simpler to implement into current hardware.

Learning
Its great to be a shareholder.
Yes of course, was that a rhetorical question Learning? 😛

Mercedes did say, that they'd developed "software and hardware" systems, using AKIDA.

Software, isn't going to give the same efficiency as hardware, but not relevant in ICE vehicles.
 
  • Like
  • Fire
  • Love
Reactions: 9 users

Labsy

Regular
"Also, Mercedes has been pretty clear that AKIDA is only being implemented in the EQS"

Have they stated, that it would only be in the EQS?

AKIDA implemented in the MBUX system, did a whole lot more, than "just" improve efficiency.
Increased "naturalness" and reponsiveness of the system, were noted (from memory).

The latest iteration of the MBUX, just won an award or something, for how much it had improved.

In my opinion, with Mercedes drive, to always have the best.
The only reason they would not be implementing the AKIDA upgraded version now, would be because the chips (possiblly designed through MegaChips) hadn't been produced yet..
Yes, but I think Mercedes are making their own chips from memory? So may take some time.
 
  • Like
Reactions: 3 users

Diogenese

Top 20
Just a question to the more informed.

Is its possible to implement Akida software first?
We know the Akida hardware will take time to implement, but the software is simpler to implement into current hardware.

Learning
Its great to be a shareholder.
The software is always implemented first.

The MetaTF/ADE is provided so the customer can run a simulation of Akida on a von Neumann computer to see how it will perform in the customer's specific application.

It would be used to determine how many NPU/nodes are required and to see how it worked with the chosen image/dictionary/smell ... library.

The software implementation is much slower and more power hungry than the Akida SoC.
 
  • Like
  • Love
  • Fire
Reactions: 27 users

Slade

Top 20
There are many different uses for Akida in Mercedes. They don’t all have to be implemented at the same time. But since Mercedes has discovered that Akida can save power and help EVs travel further I would suspect that they are working flat out to bring this technology to their vehicles. I also expect that Akida will show up in Valeo’s 3rd generation LiDAR system in 2023. My money is still on Renesas being our first source of revenue. Maybe even a trickle coming through in the next 4C. But definitely in the last quarter.
 
  • Like
  • Fire
  • Love
Reactions: 24 users
Just a question to the more informed.

Is its possible to implement Akida software first?
We know the Akida hardware will take time to implement, but the software is simpler to implement into current hardware.

Learning
Its great to be a shareholder.
The AKD1000 and the IP are processor agnostic so just relying upon logic I would think this would also apply to the software.

On the present debate about Mercedes I think you need to have regard to the fact that Hey Mercedes has been much criticised since it was introduced in 2018 and there have been multiple upgrades but from what I have read it seems to be more a criticism of what is at the back end than a criticism of the actual ability to understand the owners voice.

I therefore think it is much more than just an improvement in keyword spotting that Mercedes is implementing with AKIDA.

While they may get ahead in some areas of development I personally lean towards it being a whole system release 2024.

I would love an earlier release and do not discount it could happen and I even allow for it being for some Ai sensor about which we presently have no idea. After all in cabin monitoring of children to prevent them being left in vehicles is to be made compulsory across many different countries 2023.

My opinion only DYOR
FF

AKIDA BALLISTA
 
  • Like
  • Fire
  • Love
Reactions: 23 users
Yes, but I think Mercedes are making their own chips from memory? So may take some time.

Yes, but I think Mercedes are making their own chips from memory? So may take some time.
"Mercedes engineers worked with California-based artificial-intelligence developer BrainChip to create systems based on the company’s Akida hardware and software. Among other things, the technology makes the “Hey, Mercedes” voice control system in the EQXX five to ten times more efficient than conventional voice control"
 
  • Like
  • Love
Reactions: 23 users

Slade

Top 20
When Sean our CEO highlighted the importance of honoring NDAs and said watch the financials instead, I don’t believe he meant start watching them in a years time.
 
  • Like
  • Love
  • Fire
Reactions: 39 users

Serengeti

Regular
  • Like
  • Fire
Reactions: 8 users
The software is always implemented first.

The MetaTF/ADE is provided so the customer can run a simulation of Akida on a von Neumann computer to see how it will perform in the customer's specific application.

It would be used to determine how many NPU/nodes are required and to see how it worked with the chosen image/dictionary/smell ... library.

The software implementation is much slower and more power hungry than the Akida SoC.
Don’t forget though as brilliant as the hardware is the software ADE still produced STATE OF THE ART PERFORMANCE WITH THE AERO EXPERIMENT.

So who is to say that state of the art is not good enough for some customers?

My opinion only DYOR
FF

AKIDA BALLISTA
 
  • Like
  • Love
Reactions: 14 users
"Mercedes engineers worked with California-based artificial-intelligence developer BrainChip to create systems based on the company’s Akida hardware and software. Among other things, the technology makes the “Hey, Mercedes” voice control system in the EQXX five to ten times more efficient than conventional voice control"
Mercedes, wouldn't have said that they'd also developed software systems, if that was merely "part" of the development cycle.

Brainchip Accelerator wasn't slow @Diogenese, just too novel as a software solution at the time..

But we are just picking over words and phrases here..
 
  • Like
  • Love
  • Fire
Reactions: 9 users

Diogenese

Top 20
Don’t forget though as brilliant as the hardware is the software ADE still produced STATE OF THE ART PERFORMANCE WITH THE AERO EXPERIMENT.

So who is to say that state of the art is not good enough for some customers?

My opinion only DYOR
FF

AKIDA BALLISTA
Yes. We should not forget the award winning BrainChip Studio software which ran on von Neumann processors. The current MetaTF/ADE could be adapted to do those sorts of jobs where power is not at a premium.

https://brainchip.com/brainchip-stu...panel-of-independent-judges-brainchip-220817/

BrainChip Studio Awarded New Product of the Year for Video Analytics by Security Today’s Panel of Independent Judges 22.08.2017 AI-powered software is easily deployed in existing infrastructure, where it rapidly searches archived video or live streams to identify objects or faces ALISO VIEJO, CA — (Marketwired) — 08/22/17 — BrainChip Holdings Ltd. (ASX: BRN) (“BrainChip” […]


How cool is it to be able to shelve the 2017 award-winning AI software product to switch to the world's best AI SoC in a couple of years?

PS: I'll bet the BRN team look back on those years with unalloyed pride and satisfaction.
 
  • Like
  • Love
  • Fire
Reactions: 24 users

HopalongPetrovski

I'm Spartacus!
Mercedes, wouldn't have said that they'd also developed software systems, if that was merely "part" of the development cycle.

Brainchip Accelerator wasn't slow @Diogenese, just too novel as a software solution at the time..

But we are just picking over words and phrases here..
images-1.jpeg
 
  • Haha
  • Like
Reactions: 11 users

Diogenese

Top 20
Mercedes, wouldn't have said that they'd also developed software systems, if that was merely "part" of the development cycle.

Brainchip Accelerator wasn't slow @Diogenese, just too novel as a software solution at the time..

But we are just picking over words and phrases here..
Hi DB,

BrainChip Accelerator was hardware which enhanced the performance of BrainChip Studio software.

https://brainchip.com/brainchip-introduces-worlds-first-commercial-hardware-acceleration-of-neuromorphic-computing-brainchip-120917/#:~:text=BrainChip Accelerator is an 8-lane, PCI-Express add-in card,channels of a system to 16 per card.

BrainChip Holdings Ltd. announced the release of its hardware acceleration board, BrainChip Accelerator. SAN FRANCISCO, CA — (Marketwired) — 09/12/17 — BrainChip Holdings Ltd. (ASX: BRN) Enables 16 channels of simultaneous video processing; Provides a low power, up to 6x speed boost to BrainChip Studio’s CPU-based Artificial Intelligence Software for Object Recognition; 7x more efficient.

While Accelerator blew their socks off in 2017, Akida is much faster and uses much less power.

Akida uses software (with the ARM Cortex processor) to configure the nodes (organize the layers and weights) but it does not use software in performing object classification.
 
Last edited:
  • Like
  • Fire
  • Love
Reactions: 18 users

alwaysgreen

Top 20
Hi all,
Interesting article, thought I would share.
I wonder if Akida is inside the Ford test vehicles?
The pessimist in me worries that Ford has decided not to use Akida for self driving.

The optimist in me thinks that Argo is likely utilising AKIDA (they have stated in the past that their systems are "continually learning"). Which would have us in bed with not just Ford, but also VW Group.

Fingers crossed it's the latter!
 
Last edited:
  • Like
  • Fire
Reactions: 12 users
Hi DB,

BrainChip Accelerator was hardware which enhanced the performance of BrainChip Studio software.

https://brainchip.com/brainchip-introduces-worlds-first-commercial-hardware-acceleration-of-neuromorphic-computing-brainchip-120917/#:~:text=BrainChip Accelerator is an 8-lane, PCI-Express add-in card,channels of a system to 16 per card.

BrainChip Holdings Ltd. announced the release of its hardware acceleration board, BrainChip Accelerator. SAN FRANCISCO, CA — (Marketwired) — 09/12/17 — BrainChip Holdings Ltd. (ASX: BRN) Enables 16 channels of simultaneous video processing; Provides a low power, up to 6x speed boost to BrainChip Studio’s CPU-based Artificial Intelligence Software for Object Recognition; 7x more efficient.

While Accelerator blew their socks off in 2017, Akida is much faster and uses much less power.

Akida uses software (with the ARM Cortex processor) to configure the nodes (organize the layers and weights) but it does not use software in performing object classification.
_20220708_131959.JPG
 
  • Like
  • Haha
Reactions: 8 users

Rodney

Regular
Shorts sitting on 90 mil + will hit 100 mil next week if the average keeps continuing. Might be there already as the 90 mil is 4 days old. I can’t remember BRN ever being this high in shorts and should make One hell of a short rally if news drops. Are they starting to get desperate as there is not a lot of volume and we know how quick it can rise on volume. BRN just need to light the fire.
 
  • Like
  • Fire
  • Wow
Reactions: 29 users
F

Filobeddo

Guest
Shorts sitting on 90 mil + will hit 100 mil next week if the average keeps continuing. Might be there already as the 90 mil is 4 days old. I can’t remember BRN ever being this high in shorts and should make One hell of a short rally if news drops. Are they starting to get desperate as there is not a lot of volume and we know how quick it can rise on volume. BRN just need to light the fire.

View attachment 10904
 

Attachments

  • 06865E9F-0800-4BF6-B040-BB57272D7673.gif
    06865E9F-0800-4BF6-B040-BB57272D7673.gif
    3.3 MB · Views: 51
  • Like
  • Fire
  • Haha
Reactions: 5 users
The pessimist in me worries that Ford has decided not to use Akida for self driving.

The optimist in me thinks that Argo is likely utilising AKIDA (they have stated in the past that their systems are "continually learning". Which would have us in bed with not just ford, but also VW Group.

Fingers crossed it's the latter!
Here is something to drive a stake through the heart of your pessimist so that it never again raises its very, very ugly head.

In the Rob Telson and Jerome Nadel podcast it was stated there are 70 sensors in an automobile.

If you process these 70 sensors with non AKIDA edge Ai best view outcome is it will cost you 1 to 1.5 watts so you will have a total draw of 70 to 105 watts.

Mercedes’ Benz stated that AKIDA technology was 5 to 10 times more efficient than comparable products so the 70 to 105 watts becomes 14 to 21 watts or 7 to 10.5 watts or some number in between.

You then achieve the additional saving in the GPU running the whole system because it is not receiving all data but the meta data which does not need preprocessing as this has been done on sensor already by AKIDA.

It is incontestable that driving range in an EV can only be extended by ensuring the maximum percentage of the battery power is reserved for the driving wheels.

Blind Freddie knows that the sort of energy savings that AKIDA brings to the EV platform cannot be ignored.

I personally along with Blind Freddie have complete confidence that AKIDA technology will dominate in the EV space on power savings alone and this is before you add in all of its other unique patent protected features.

The word ‘UBIQUITOUS’ is not being thrown around for nothing.

My opinion only DYOR
FF

AKIDA BALLISTA
 
Last edited:
  • Like
  • Love
  • Fire
Reactions: 66 users
Top Bottom