BRN Discussion Ongoing

[Diogenese]

This is how I am looking at it, waiting for @Diogenese response

Hi Jesse,

I responded last evening:

• #67,177

@Diogenese ??? Now Brainchip is Edge Impulses only strategic IP partner…..what the hell does this mean? Anybody?
The way I see it is the less sophisticated customers of Brainchip and edge impulse will use Brainchips IP.

Also note that ARM and Brainchip are on the right hand side, I assume due to both being an IP provider.
View attachment 46283

Hi Jesse,

Zack was talking about the enormous amount of data Nvidia had in its AI model libraries, and about Edge Impulse adapting Nvidia model libraries for the different types of processors, and the graph indicates the size of model for the different processors.

This may be linked to Akida 2 adopting 8-bit weights.

The graph also shows that the Akida models can be much smaller than the GPU or CPU models while still performing at the level of a GPU (or better).

The intriguing thing is that ARM is boxed with Akida? - probly because Akida 1 uses Cortex?

 

[equanimous]

Tata Technologies Achieves 18% Reduction in Power Consumption Using AI/ML - Techiexpert.com

In a pioneering initiative, Tata Technologies has harnessed the capabilities of artificial intelligence and machine learning to revolutionize its

www.techiexpert.com

 

[Earlyrelease]

[QUOTE="Diogenese, post: 375757, member:
Dodgy Knees
If you look one further line items down what interested me in Apogee in addition to what you have highlighted is their Apollo program -“advancing the use of AI in support of image processing for intelligence purposes.

Now what could Akida possibly do to help this goal..????

Nah he moved over east because he didn’t like the sun.

 

[Quiltman]

Tata Technologies Achieves 18% Reduction in Power Consumption Using AI/ML - Techiexpert.com

In a pioneering initiative, Tata Technologies has harnessed the capabilities of artificial intelligence and machine learning to revolutionize its

www.techiexpert.com

This part is particularly enticing. Thanks for sharing.

they are spending a lot of money to make a special hub for the new technology

Particularly as industrial applications is a focus

Tata Elxsi’s partnership with BrainChip will be driving Akida™ technology into medical devices and industrial applications

 

[7für7]

Everything's great, guys and gals! Thanks for the info! Nevertheless, I'm here waiting for official announcements! I don't want to peek into a restaurant, see them baking crispy turkeys, get excited, and then watch them devour others! We need news from BrainChip!

 

[Diogenese]

Hi Jesse,

I responded last evening:

• #67,177

Hi Jesse,

Zack was talking about the enormous amount of data Nvidia had in its AI model libraries, and about Edge Impulse adapting Nvidia model libraries for the different types of processors, and the graph indicates the size of model for the different processors.

This may be linked to Akida 2 adopting 8-bit weights.

The graph also shows that the Akida models can be much smaller than the GPU or CPU models while still performing at the level of a GPU (or better).

The intriguing thing is that ARM is boxed with Akida? - probly because Akida 1 uses Cortex?

What I should have added was that if you insert an imaginary vertical axis labeled "model size" and an imaginary horizontal axis labeled "performance", the graph can be interpreted as showing that Akida only needs a very small model library compared to GPU for similar or better performance.

 

[jtardif999]

Great posts of late ... especially from Chapman89 ......
However, I'm very concerned about the next AGM's potential "Strike 2" scenario playing out in the event that a sustained positive revenue stream is not forthcoming by then ....... I am presuming that without such a revenue stream eventuating or even a lack of IP contracts being entered into that there will be a lot of negative Co and management sentiment, thus placing the BOD and management generally on notice as to their ongoing positions within the Co ................

Personally I strongly hope that the CEO and BOD are well and truely well aware of this potential critical state of affairs and will have this all well under control before the next AGM.

Strike 2 whilst being disruptive would just mean the board would have to be re-elected. If there are no better alternatives than the current board and ceo then they would all be re-elected. No actual drama, just a lot of inconvenience imo.

 

[Dhm]

What I should have added was that if you insert an imaginary vertical axis labeled "model size" and an imaginary horizontal axis labeled "performance", the graph can be interpreted as showing that Akida only needs a very small model library compared to GPU for similar or better performance.

How about this @Diogenese

 

[buena suerte :-)]

Hi Guys & Gals ...heaps of info over at

(Thread) I think @Quiltman could do with some Chippers reading some great info posted includes FF saying Hi!!! ...over & out

 

[jtardif999]

Well his answers weren't that great. He's pushing RealityAI ( a subsidiary of Renesas) - which is cloud based. He doesn't like 8 bit compute, preferring floating point. I think he's doing a sales pitch for their RealityAI product which is in competition to Akida.

But great you asked, and funny how he knew both your questions were linked.

I'm ecstatic with our Akida 2 announcement. We needed it because we said we would make this announcement. I don't fault our management in any way at all. The Akida 2 is what clients have asked for, and it's been delivered!

I do question Renesas' commitment to Akida as they have spent $$$ on their own competing products and only ever licensed the 2 nodes. I don't think it is by chance that we went with MegaChips to produce the Akida 1500 chip.

Rubbish post; firstly Reality AI are a software company not in anyway a competitor to Akida. I think from memory there’s a possibility they could support some product development involving Akida. Secondly Renesas have their own DRP reconfigurable AI processor which they’ve spent many years developing and want to see some pay dirt for it for sure, but the DRP processor is limited in the applications it is suited to and this is why Renesas licenced the 2 Akida nodes - so they could sell MCUs with a smaller silicon footprint where they couldn’t sell DRP.

 

[Bravo]

Not BRN related but great news for this little guy who was just declared the winner of the Australian Air Guitar Championships!

 

[Fastback6666]

I don't know, folks, it always looks great at first glance. However, I must admit that I approach this with a clear head. I think to myself, 'Just because they've partnered with BrainChip doesn't necessarily mean they'll exclusively rely on it.' All these partners have different agreements with other companies in this field, and it would be a milestone if everyone always turned to us. Personally, I'm not holding onto that too tightly. Otherwise, you'll end up disappointed later….like for example the new Mercedes CLA, that has a water-cooled MBUX processor... If they had taken Akida, they wouldn't have to cool it separately It would only operate when needed... with less energy... but well, if they want to consume additional energy for cooling

But Brainchip are not focusing as being the core (CPU or GPU) processor solution really at this point. Over time this may change.

Brainchip really are more focused on solving sensors and extreme edge problems which would be the all the smaller sensors and ancillary functions on the car.

 

[7für7]

But Brainchip are not focusing as being the core (CPU or GPU) processor solution really at this point. Over time this may change.

Brainchip really are more focused on solving sensors and extreme edge problems which would be the all the smaller sensors and ancillary functions on the car.

Yes, I know..I am invested around 2 1/2 years in brainchip… it was just an example! Brainchip will solve a lot of problems. But how I said, nothing counts without a nice announcement from brainchip.

 

[Iseki]

Rubbish post; firstly Reality AI are a software company not in anyway a competitor to Akida. I think from memory there’s a possibility they could support some product development involving Akida. Secondly Renesas have their own DRP reconfigurable AI processor which they’ve spent many years developing and want to see some pay dirt for it for sure, but the DRP processor is limited in the applications it is suited to and this is why Renesas licenced the 2 Akida nodes - so they could sell MCUs with a smaller silicon footprint where they couldn’t sell DRP.

Renesas acquired Reality AI.

 

[chapman89]

Renesas acquired Reality AI.

Reality AI commented on the akida 2nd generation back in March.
Think it’s safe to say we are working with them, not against them!

“We see a growing number of predictive industrial (including HVAC, motor control) or automotive (including fleet maintenance), building automation, remote digital health equipment and other AIoT applications use complex models with minimal impact to product BOM and need faster real-time performance at the Edge” said Nalin Balan, Head of Business Development at Reality ai, a Renesas company. “BrainChip’s ability to efficiently handle streaming high frequency signal data, vision, and other advanced models at the edge can radically improve scale and timely delivery of intelligent services.”

 

[Iseki]

So summary is that it appears a fair plus possibility AKIDA will end up in NVIDIA products to the final user via Edge Impulse and maybe TAT???
Correct me if i am wrong.

Other way around, I think.
Massive NVIDIA zoo models can be run on Akida based chips.
This is better for us in that OEMs currently working with NVIDIA can swap to an AKIDA based chip once their product has been developed.

 

[Iseki]

Reality AI commented on the akida 2nd generation back in March.
Think it’s safe to say we are working with them, not against them!

“We see a growing number of predictive industrial (including HVAC, motor control) or automotive (including fleet maintenance), building automation, remote digital health equipment and other AIoT applications use complex models with minimal impact to product BOM and need faster real-time performance at the Edge” said Nalin Balan, Head of Business Development at Reality ai, a Renesas company. “BrainChip’s ability to efficiently handle streaming high frequency signal data, vision, and other advanced models at the edge can radically improve scale and timely delivery of intelligent services.”

That would be brilliant. Now where did we get up to with that M85+Akida chip?

 

[IloveLamp]

Arm Tech Talk from Plumerai: Accelerating People Detection with Arm Helium vector extensions

Register for our upcoming Arm Tech Talks at https://www.arm.com/techtalks In this talk, Plumerai will showcase their people detection demo on an Arm Cortex-M...

m.youtube.com

 

[IloveLamp]

#aimlsys #edgecomputing #neuromorphic #aimlsys #sai4e #workshop #bangalore #inventingforimpact | Arijit Mukherjee

Tata Consultancy Services - Research is organising the 1st International Workshop on Sustainable AI for Edge as part of #AIMLSys 2023 where #EdgeComputing meets #Neuromorphic Four technical papers - selected after double blind reviews. Keynote speeches: ▪️Ghanashyam (Shyam) Prabhakar...

www.linkedin.com

Registration - The Third International Conference on AI ML Systems

**All the fees are excluding GST. Additional GST@18% will be added at the payment page.

www.aimlsystems.org

 

[robsmark]

How's about I get you both a beer on one of our super yachts.... easy fellas ...

I accept