Fact Finder
Top 20
Your making out AKIDA to be like the try hard kid in my maths class who always had his hand up to answer questions and everyone hated.
BRN Discussion Ongoing
- Thread starter TechGirl
- Start date
NXP are the next most versatile, but they are no competition - they would not know a neuromorph if they fell over it:
https://www.digikey.com.au/en/product-highlight/n/nxp-semi/i-mx-8m-nano-ultralite
Fact Finder
Top 20
If you have a few spare minutes this little article is very illuminating when you know as we do that Brainchip has the first commercial neuromorphic IP and reference chip capable of:
1. On chip one shot, few shot and incremental learning;
2. On chip multi sensor processing - evidence Nviso working towards having AKIDA 1.0 process all 20 odd Apps on chip, and
3. On chip multi sensor fusion to create actionable instructions.
As the concluding paragraph states it is great to think ahead as Peter van der Made and Anil Mankar have by at least three years over everyone else in the World protected by patents.
My opinion only DYOR
FF
AKIDA BALLISTA
1. On chip one shot, few shot and incremental learning;
2. On chip multi sensor processing - evidence Nviso working towards having AKIDA 1.0 process all 20 odd Apps on chip, and
3. On chip multi sensor fusion to create actionable instructions.
AI Uses Artificial Sleep To Learn New Task Without Forgetting The Last - Public News Time
Many AIs can only become good at one task, forgetting everything they know if they learn another. A form of artificial sleep could help stop this from
publicnewstime.com
As the concluding paragraph states it is great to think ahead as Peter van der Made and Anil Mankar have by at least three years over everyone else in the World protected by patents.
My opinion only DYOR
FF
AKIDA BALLISTA
wilzy123
Founding Member
FF May I translate for you
When you search for an ecosystem partner on the ARM website using the search "by relevance"
https://www.arm.com/partners/catalog/results#sort=relevancy
Brainchip is offered by Arm (Therefore potentially considered by ARM) as the first preference for that chosen solution.
Its not alpah order so the decision to have a particular company first in the offereings seems based on another reason.
In Germany being FIRST is seen to be an important statement
Mercedes knows this
From what I can tell based on my 5 minutes worth of poking around the source code of that page, each item in the result set receives a score. When you sort by 'relevance', Brainchip comes up top because it has the highest 'score'.
The site is using a service called coveo search, as part of the sitecore CMS. According to coveo documentation, The relevance score is a combination of the index ranking algorithm in action during the index ranking phases, and other relevance modifiers such as query ranking expressions (QRE) and query ranking functions.
While the index ranking is supposedly algorithmic, I am not sure what it is actually indexing in order to produce a score.
In any case - we are obviously very 'relevant'
Sirod69
bavarian girl ;-)
Rob Telson likes it and I remember, we spoke about onsemi not long ago:
Sometimes business at onsemi feels like it’s moving 200mph, and next year it will be. Our partnership with Arrow McLaren SP will get our innovative technology on the track, in the groove…and hopefully to the podium.
#onsemiontrack #weAREonsemi #IndyCar
PHOENIX – Nov. 9, 2022 – onsemi (Nasdaq: ON), a leader in intelligent power and sensing technologies, today announced that it joined McLaren Racing as an Official Partner of the Arrow McLaren SP Team and its No. 6 Chevrolet, piloted by Felix Rosenqvist, for the 2023 NTT INDYCAR SERIES season. As a leading semiconductor manufacturer with a global supply chain, onsemi delivers intelligent technology solutions to help customers create products focused on sustainability, including electric vehicles and advanced safety features, to increase energy efficiency.
Arrow McLaren SP relies on onsemi components in critical hardware on its race cars, such as steering wheels, enabling a more efficient design, without compromising power and safety. As McLaren looks forward to the new hybrid powertrain, onsemi will continue to be a key partner, as it builds a more sustainable racing environment.
“We’re delighted that onsemi will be joining us to lead the identity of Felix’s race car in 2023,” said Matt Dennington, executive director, Partnerships & Accelerator, McLaren Racing. “This partnership is crucial to our team, with onsemi’s technology ingrained in the Arrow McLaren SP race cars. Like McLaren, onsemi is determined to make a positive impact beyond its sector, making this a transformational union of brands both on and off the racetrack. We look forward to a successful partnership and joining forces in pursuit of the IndyCar Championship together.”
The partnership unites two brands in alignment with their pioneering spirits. McLaren has a history of pushing the limits by competing in different race series and applying its technological innovation outside of motorsport, while onsemi thrives on being a disruptive semiconductor company, surpassing competition through a broad, solutions-based portfolio and premium experiences across industries.
onsemi will be the car title partner for Felix’s No. 6 Arrow McLaren SP Chevrolet for a selection of races in 2023, including the Grand Prix of Long Beach, Chevrolet Detroit Grand Prix, Grand Prix at Road America, and the Bommarito Automotive Group 500. The onsemi brand will be represented on all three Arrow McLaren SP Chevrolet cars and team kit throughout the full 2023 season.
"Our partnership with Arrow McLaren SP, along with Felix behind the wheel, is a prime opportunity to reinforce onsemi’s commitment to automotive safety and efficiency, while demonstrating our superior power management and hyper-performance capabilities,” said Hassane El-Khoury, president and chief executive officer, onsemi. ”Over the years, not only has racing focused on producing the fastest and best performance for the racing teams, it has also been a catalyst for disruptive, breakthrough technologies in safety and efficiency – many of which have made their way into the consumer market. This platform will enable us to push the next generation of advancements in sustainable automotive solutions and boost consumer adoption.”
www.onsemi.com
Sometimes business at onsemi feels like it’s moving 200mph, and next year it will be. Our partnership with Arrow McLaren SP will get our innovative technology on the track, in the groove…and hopefully to the podium.
#onsemiontrack #weAREonsemi #IndyCar
onsemi joins Arrow McLaren SP as a Primary Partner of No. 6 IndyCar piloted by Felix Rosenqvist
PHOENIX – Nov. 9, 2022 – onsemi (Nasdaq: ON), a leader in intelligent power and sensing technologies, today announced that it joined McLaren Racing as an Official Partner of the Arrow McLaren SP Team and its No. 6 Chevrolet, piloted by Felix Rosenqvist, for the 2023 NTT INDYCAR SERIES season. As a leading semiconductor manufacturer with a global supply chain, onsemi delivers intelligent technology solutions to help customers create products focused on sustainability, including electric vehicles and advanced safety features, to increase energy efficiency.
Arrow McLaren SP relies on onsemi components in critical hardware on its race cars, such as steering wheels, enabling a more efficient design, without compromising power and safety. As McLaren looks forward to the new hybrid powertrain, onsemi will continue to be a key partner, as it builds a more sustainable racing environment.
“We’re delighted that onsemi will be joining us to lead the identity of Felix’s race car in 2023,” said Matt Dennington, executive director, Partnerships & Accelerator, McLaren Racing. “This partnership is crucial to our team, with onsemi’s technology ingrained in the Arrow McLaren SP race cars. Like McLaren, onsemi is determined to make a positive impact beyond its sector, making this a transformational union of brands both on and off the racetrack. We look forward to a successful partnership and joining forces in pursuit of the IndyCar Championship together.”
The partnership unites two brands in alignment with their pioneering spirits. McLaren has a history of pushing the limits by competing in different race series and applying its technological innovation outside of motorsport, while onsemi thrives on being a disruptive semiconductor company, surpassing competition through a broad, solutions-based portfolio and premium experiences across industries.
onsemi will be the car title partner for Felix’s No. 6 Arrow McLaren SP Chevrolet for a selection of races in 2023, including the Grand Prix of Long Beach, Chevrolet Detroit Grand Prix, Grand Prix at Road America, and the Bommarito Automotive Group 500. The onsemi brand will be represented on all three Arrow McLaren SP Chevrolet cars and team kit throughout the full 2023 season.
"Our partnership with Arrow McLaren SP, along with Felix behind the wheel, is a prime opportunity to reinforce onsemi’s commitment to automotive safety and efficiency, while demonstrating our superior power management and hyper-performance capabilities,” said Hassane El-Khoury, president and chief executive officer, onsemi. ”Over the years, not only has racing focused on producing the fastest and best performance for the racing teams, it has also been a catalyst for disruptive, breakthrough technologies in safety and efficiency – many of which have made their way into the consumer market. This platform will enable us to push the next generation of advancements in sustainable automotive solutions and boost consumer adoption.”
Intelligent Power and Sensing Technologies | onsemi
Just on the point about 1 Alkida performing several different functions, that would require a separate model library for each function.If you have a few spare minutes this little article is very illuminating when you know as we do that Brainchip has the first commercial neuromorphic IP and reference chip capable of:
1. On chip one shot, few shot and incremental learning;
2. On chip multi sensor processing - evidence Nviso working towards having AKIDA 1.0 process all 20 odd Apps on chip, and
3. On chip multi sensor fusion to create actionable instructions.
AI Uses Artificial Sleep To Learn New Task Without Forgetting The Last - Public News Time
Many AIs can only become good at one task, forgetting everything they know if they learn another. A form of artificial sleep could help stop this frompublicnewstime.com
As the concluding paragraph states it is great to think ahead as Peter van der Made and Anil Mankar have by at least three years over everyone else in the World protected by patents.
My opinion only DYOR
FF
AKIDA BALLISTA
Our software engineers have assembled and are assembling different model libraries. So, as part of our business model, I assume that we license out these model libraries in a similar way to the way computer software is licensed (annual fees for upgrades etc.), so we need to add this into any considerations about how much BrainChip will earn from each Akida IP licence.
One of our patents covers the process of uploading newly learned images/sounds etc to a central server where the newly learnt examples can be added to the libraries and distributed the the appropriate Akida installations.
The model library licence fees could become our major source of income.
PS: There are a number of CNN model libraries available, but they need to be adapted for use with Akida.
Last edited:
Fact Finder
Top 20
Many thanks @WilzyFrom what I can tell based on my 5 minutes worth of poking around the source code of that page, each item in the result set receives a score. When you sort by 'relevance', Brainchip comes up top because it has the highest 'score'.
The site is using a service called coveo search, as part of the sitecore CMS. According to coveo documentation, The relevance score is a combination of the index ranking algorithm in action during the index ranking phases, and other relevance modifiers such as query ranking expressions (QRE) and query ranking functions.
While the index ranking is supposedly algorithmic, I am not sure what it is actually indexing in order to produce a score.
In any case - we are obviously very 'relevant'
AND I just love your last line:
“In any case - we are obviously very 'relevant'
AND that’s it in a nutshell.
My opinion only DYOR
FF
AKIDA BALLISTA
Top notch sleuthing Wilzy,From what I can tell based on my 5 minutes worth of poking around the source code of that page, each item in the result set receives a score. When you sort by 'relevance', Brainchip comes up top because it has the highest 'score'.
The site is using a service called coveo search, as part of the sitecore CMS. According to coveo documentation, The relevance score is a combination of the index ranking algorithm in action during the index ranking phases, and other relevance modifiers such as query ranking expressions (QRE) and query ranking functions.
While the index ranking is supposedly algorithmic, I am not sure what it is actually indexing in order to produce a score.
In any case - we are obviously very 'relevant'
So does that mean that, each time the 1000 eyes searches the ARM partners page for "BrainChip", our score goes up?
Just one little daily chore for the 1000 eyes ...
wilzy123
Founding Member
Haha, maybe. It all depends on what 'ranking factors' are used in determining a relevance score as part of that particular search engine.
Actual use of the page might be one - but I cannot see that far into their configuration.
We could see if the score changes over time. BRN has a score of 3732, while the 'next best result' (NXP) has a score of 2761. All of this information is visible from within the response to the AJAX request that the page makes each time it looks for results: https://www.arm.com/coveo/rest/search/v2?sitecoreItemUri=sitecore://web/{7502997D-E820-488F-8264-B8BA0B39CBA6}?lang=en&ver=3&siteName=arm-redesign-website
The score will likely change each time the pool gets re-indexed (I.e. to keep the 'relevance' fresh) - I suspect this is done maybe once a week for a site like this that doesn't change that much.
Last edited:
Fact Finder
Top 20
The new black in computing at least according to this lengthy article which I have scrolled through on Earth and in Space is ARD - Analysis Ready Data:
It is worth having just a little peak at it, if for no other reason than, to realise that there are actually people in this world whose brains work in these mysterious ways.
However after you master the convoluted writing style effectively what they are trying to say is that sensors need to be made smart in five million words or less with about 500 supporting references.
My takeaway is that:
1. Valeo Brainchip Scala AKIDA Lidar with 3 D point cloud would meet ADR;
2. Prophesee Brainchip powered event based vision sensors would meet ADR; and
3. Nviso AKIDA Brainchip powered human monitoring Apps for ADAS and medical applications would meet ADR.
All of which reminds me of a song from my youth:
‘What the world needs now is AKIDA ADR it’s the only thing that there is just too little of….’ - Sorry Burt.
My opinion only DYOR
FF
AKIDA BALLISTA
It is worth having just a little peak at it, if for no other reason than, to realise that there are actually people in this world whose brains work in these mysterious ways.
However after you master the convoluted writing style effectively what they are trying to say is that sensors need to be made smart in five million words or less with about 500 supporting references.
My takeaway is that:
1. Valeo Brainchip Scala AKIDA Lidar with 3 D point cloud would meet ADR;
2. Prophesee Brainchip powered event based vision sensors would meet ADR; and
3. Nviso AKIDA Brainchip powered human monitoring Apps for ADAS and medical applications would meet ADR.
All of which reminds me of a song from my youth:
‘What the world needs now is AKIDA ADR it’s the only thing that there is just too little of….’ - Sorry Burt.
My opinion only DYOR
FF
AKIDA BALLISTA
Last edited:
Never have I been so bethumped with words ... he cudgels our ears ... he gives the bastinado with his tongue ...
Fact Finder
Top 20
I have no idea really but on the ARM website:
1. to be rating close to 1,000 points higher
2. than the next highest ARM partner
3. when that partner is NXP which according to Wiki “is a Dutch semiconductor designer and manufacturer with headquarters in Eindhoven, Netherlands.
A company that employs approximately 31,000 people in more than 30 countries.
With reported revenue of $11.06 billion in 2021.
Traded as: Nasdaq: NXPI; NASDAQ-100 component; S&P 500 component
Revenue: US$11.063 billion (2021)
4. IS THIS NOT JUST LIKE THE HUGESTEST PIECE OF FACTUAL INFORMATION LIKE FOREVER, and,
Perhaps why the CEO Sean Hehir in the last 4C Report stated:
“We are seeing the greatest amount of sales activity and engagement in the Company’s history”
My opinion only DYOR
FF
AKIDA BALLISTA
Fact Finder
Top 20
And if it is the hugestest ever FACT then we have @Baneino to thank for kicking off the 1,000 Eye research machine.
My opinion only DYOR
FF
AKIDA BALLISTA
chapman89
Founding Member
Without even typing anything in the search bar we sit at the top as well….5 above the largest company in the world…APPLE
Attachments
wilzy123
Founding Member
Who can check this…DNN…new competitor?
www.linkedin.com
Edge Impulse on LinkedIn: Ultra-low Power DNN Accelerators for IoT: Resource Characterization of the…
In this work, researchers at Nokia Bell Labs closely examined the ultra-low-power MAX78000 AI accelerator by Maxim Integrated. The authors characterized the…
Fact Finder
Top 20
Not yet.Who can check this…DNN…new competitor?
Edge Impulse on LinkedIn: Ultra-low Power DNN Accelerators for IoT: Resource Characterization of the…
In this work, researchers at Nokia Bell Labs closely examined the ultra-low-power MAX78000 AI accelerator by Maxim Integrated. The authors characterized the…
It is only working with CNN models and is purely an accelerator and has the following issues:
“Key takeaways:
The results in this subsection imply that it is important to use the minimum number of groups with consecutive processors for runtime performance optimization.
While the latency overhead seems small in absolute terms (≈ 200𝜇s), it adds up quickly for models with many layers and results in significant penalties in terms of inference time, and consequently energy consumption.
The optimal processor placement is still an open problem given that automatic tools are not provided.
We leave this for future work.
5 CONCLUSION
In this paper, we conducted a variety of benchmark studies to char- acterize the resource and performance of the ultra-low power DNN accelerator, MAX78000.
First, we analyzed the operational latency, power consumption, and memory footprint of five DNN models with various sizes and architecture.
Second, we further investigated the system implications in terms of the model architecture and convolutional processor selection in order to maximize the accel- eration.
Beyond the numbers, our benchmark study further offers meaningful insights for the development of on-device AI systems on ultra-low power, tiny-scale AI accelerators”
My opinion only DYOR
FF
AKIDA BALLISTA
Fact Finder
Top 20
Just like every other verification engineer I have ever known. Bone idle can’t even be bothered to type into a search bar. No wonder you can’t get a real job when you are happy to be just five places above APPLE.Without even typing anything in the search bar we sit at the top as well….5 above the largest company in the world…APPLE
Who can check this…DNN…new competitor?
Edge Impulse on LinkedIn: Ultra-low Power DNN Accelerators for IoT: Resource Characterization of the…
In this work, researchers at Nokia Bell Labs closely examined the ultra-low-power MAX78000 AI accelerator by Maxim Integrated. The authors characterized the…
US2021216868A1 SYSTEMS AND METHODS FOR REDUCING MEMORY REQUIREMENTS IN NEURAL NETWORKS
9 . A system for processing large amounts of neural network data, the system comprising:
a processor; and
a non-transitory computer-readable medium comprising instructions that, when executed by the processor, cause steps to be performed, the steps comprising:
determining one or more active layers in a neural network;
using the one or more active layers to process a subset of a set of input data of a first neural network layer the subset having a data size that is substantially less than the size of the set of input data;
outputting a first set of output data from the first network layer;
using first set of output data in a second neural network layer; and
outputting a second set of output data from the second network layer prior to processing all of the set of input data.
Looks like an attempt to claim quasi-asynchronous processing of CNN data on a CPU/GPU.
HopalongPetrovski
I'm Spartacus!
Similar threads
