BRN Discussion Ongoing

[krugerrands]

Looks like they are using DRP-AI3.

https://www.renesas.com/sites/default/files/RZV2H-block-chart-DRP-AI3.png
View attachment 58189

Renesas Develops New AI Accelerator for Lightweight AI Models and Embedded Processor Technology to Enable Real-Time Processing

www.renesas.com

...
It achieved up to 16 times faster processing (130 TOPS) than before the introduction of these new technologies, and world-class power efficiency (up to 23.9 TOPS/W at 0.8 V supply).
...
solve this issue, Renesas optimized its unique DRP-based AI accelerator (DRP-AI) for pruning. By analyzing how pruning pattern characteristics and a pruning method are related to recognition accuracy in typical image recognition AI models (CNN models), we identified the hardware structure of an AI accelerator that can achieve both high recognition accuracy and an efficient pruning rate, and applied it to the DRP-AI design.
...
Renesas created a prototype of a test chip with these technologies and confirmed that it achieved the world-class, highest power efficiency of 23.9 TOPS per watt at a normal power voltage of 0.8 V for the AI accelerator and operating power efficiency of 10 TOPS per watt for major AI models.
...

Anything you see that will confirm or preclude this DRP-AI3 from being Akida IP ?

 

[manny100]

One thing to remember is that several of our early engagements are joint developments rather than licences. For example, our Valeo relationship is a joint development. The development will proceed in a "commercial-in-confidence" basis until it's commercialized. A JD would not necessarily attract a licence fee, just a share of profit. So I'm hoping that Scala 3, due out soon, will include Akida.

Other JD partners include:
nViso
Edge impulse
Prophesee
SiFive
Socionext


... and of course there are the other 100+ NDAs.

Looks like the early adapters as listed in the Presentation the other day may well be the first scale users of AKIDA.
Verago, Valeo, NASA and Mercedes.
Of course there may be a few quietly working in the background.

 

[Tothemoon24]

• 
• 
  
• 
  
• https://www.design-reuse.com/news/5...vision-radar-accelerator-dsp-automotive.html?
• 
  
• 
  
• 
  
• The new Vision DSPs deliver up to 2X computer vision performance improvements for computer vision filter and NMS algorithms, as well as AI improvements for quantization and depthwise separable convolution.

The Vision 341 and Vision 331 DSPs support the Tensilica Instruction Extension (TIE) language, allowing customers to customize the instruction set. The Cadence NeuroWeave™ Software Development Kit (SDK) provides neural network support for both DSPs. In addition, the Vision 341 and Vision 331 DSPs support more than 1700 OpenCV-based vision library functions, SLAM Library, Point Cloud Library (PCL), Radar library, Nature DSP library, OpenCL and the Halide compiler for computer vision, imaging, radar and lidar applications. Both cores are automotive ready with ASIL-B hardware random faults and ASIL-D systematic fault certification.

“With the growing prevalence of vision transformers, automotive SoC and systems companies need best-in-class performance for sensor fusion applications. Increasingly, they are seeking a highly flexible single-DSP solution for computer vision, radar and lidar with AI processing,” said David Glasco, VP of R&D for the Silicon Solutions Group at Cadence. “As the leading provider of vision and radar DSP IP, we’re extremely well positioned to capitalize on this growing need by offering our customers the best of both worlds in a single DSP with optional radar acceleration for emerging 4D radar applications.”



Availability

Tensilica Vision DSPs support Cadence’s Intelligent System Design™ strategy, enabling SoC design excellence. The new Vision 331 DSP and Vision 341 DSP are available now, while the Vision 4DR accelerator will be in general availability in the second quarter of 2024. For more information on the Vision 331 DSP and Vision 341 DSP, please visit:

 

[Colorado23]

Probably not the forum all so I apologise in advance.
Anyone tried testing out superfund maximum allocation of 20% in a single stock. Will they force sell if the stock price goes to $5 as the stock now makes up 80% of your portfolio.

 

[Zedjack33]

Probably not the forum all so I apologise in advance.
Anyone tried testing out superfund maximum allocation of 20% in a single stock. Will they force sell if the stock price goes to $5 as the stock now makes up 80% of your portfolio.

Had that problem when she went over $2.

Didn’t get advice to sell.

 

[Monkeymandan]

Probably not the forum all so I apologise in advance.
Anyone tried testing out superfund maximum allocation of 20% in a single stock. Will they force sell if the stock price goes to $5 as the stock now makes up 80% of your portfolio.

I’m with Australian Super who have a 20% limit for any one stock. I have three stocks which have all bloated over the 20% due to SP increases and they don’t appear to do anything beyond warning against it.

 

[Quatrojos]

Probably not the forum all so I apologise in advance.
Anyone tried testing out superfund maximum allocation of 20% in a single stock. Will they force sell if the stock price goes to $5 as the stock now makes up 80% of your portfolio.

They haven't forced a sale of any of my BRN super shares. I'm over 30% so far...

 

[Boab]

Probably not the forum all so I apologise in advance.
Anyone tried testing out superfund maximum allocation of 20% in a single stock. Will they force sell if the stock price goes to $5 as the stock now makes up 80% of your portfolio.

Australian Super has a-limit of 20% on purchases. Now that the price has increases my percentage of BRN is over 40% of my total
So no, they wont force you to sell.

 

[Quatrojos]

They haven't forced a sale of any of my BRN super shares. I'm over 30% so far...

It'd be an interesting conversation, though:

Sir, we'd like you to sell a parcel of the shares you chose because they're performing much better than our selection...

 

[Chris B]

Probably not the forum all so I apologise in advance.
Anyone tried testing out superfund maximum allocation of 20% in a single stock. Will they force sell if the stock price goes to $5 as the stock now makes up 80% of your portfolio.

I'm in host plus that has 20% allocation and now up to near 40%... so all good... once you buy in, you can keep. Even if your stock drops out of asx 300

 

[Diogenese]

DRP AI3

AI Accelerator: DRP-AI

www.renesas.com

White paper:

https://www.renesas.com/us/en/document/whp/next-generation-highly-power-efficient-ai-accelerator-drp-ai3-10x-faster-embedded-processing?r=25422791

Looks like it works in "Spikes" just called something different. Also Pytorch which Sean mentioned Brainchip start support if for the customer(s)

It's hot and humid and still, and I'm grumpy - show me the spikes.

 

[Pom down under]

It's hot and humid and still, and I'm grumpy - show me the spikes.

 

[IloveLamp]

Extending Arm Total Design Ecosystem to Accelerate Infrastructure Innovation

Arm Total Design, a comprehensive solution for SoC design and verification, is growing, enabling partners to get to market faster, cut costs and improve results.

newsroom.arm.com

 

[Beebo]

DRP AI3

AI Accelerator: DRP-AI

www.renesas.com

White paper:

https://www.renesas.com/us/en/document/whp/next-generation-highly-power-efficient-ai-accelerator-drp-ai3-10x-faster-embedded-processing?r=25422791

Looks like it works in "Spikes" just called something different. Also Pytorch which Sean mentioned Brainchip start support if for the customer(s)

The white paper even mentions N:M.

 

[Beebo]

It's hot and humid and still, and I'm grumpy - show me the spikes.

Hi Diogenes. Please translate…do you agree or do you not agree that DRP AI3 indeed utilizes our IP?

Can’t wait to translate your reply!

 

[White Horse]

Hi All,
There is a new search engine that I learned of on THE Qube, called Perplexity.
I gave it a quick trial by asking, "Who is using Brainchip's Akida in their research."

https://www.perplexity.ai/search/Who-is-using-DxfQWL3eSmSOgg4RlmOdLQ

Now that's what I call refreshingly accurate.

 

[Diogenese]

Hi Diogenes. Please translate…do you agree or do you not agree that DRP AI3 indeed utilizes our IP?

Can’t wait to translate your reply!

Our N-of-M coding is based on the timing of activation spikes which arrive asynchronously. All spikes after the first N are discarded.

Renesas prunes the weights stored in memory in accordance with the amplitude value. They process the N largest values. N is selectable depending on required accuracy.

Now here's the rub - the timing of the activation spikes is based on the amplitude of the activation signals. Thus a similar result is obtained, but, as Ella says, "tain't what you do, it's the way that you do it!"

So, in retrospect, a similar principle underlies both techniques, but Renesas are not using our N-of-M coding in the white paper.

FN: I don't know if we use N:M to prune weights.

 

[Space Cadet]

Well, who would've thought that a stroll around the block wouldn't solve all our problems? Maybe the stock price just needs a hefty dose of mosquito-infested mulch to really take off!.. or Maybe we should all start tending to our gardens to strengthen our portfolios And who knows, maybe your Hallux valgus will become the next fitness trend in the stock market world. Either way,…… FOLKS YOU HEAR IT! Start tending your garden!!!!

Brazilian?


SC

 

[Diogenese]

Brazilian?


SC

Brazil nuts?

 

[Bravo]

Our N-of-M coding is based on the timing of activation spikes which arrive asynchronously. All spikes after the first N are discarded.

Renesas prunes the weights stored in memory in accordance with the amplitude value. They process the N largest values. N is selectable depending on required accuracy.

Now here's the rub - the timing of the activation spikes is based on the amplitude of the activation signals. Thus a similar result is obtained, but, as Ella says, "tain't what you do, it's the way that you do it!"

So, in retrospect, a similar principle underlies both techniques, but Renesas are not using our N-of-M coding in the white paper.

FN: I don't know if we use N:M to prune weights.

Maybe Renesas is doing this to lay the ground work for their new MCU with AKIDA + 22nm CMOS process with the integration of a software define radio, which is soon to be launched?