BRN Discussion Ongoing

Kachoo

Regular
In the last few years, the AI field has seen the most rapid advancement in technology ever.
Concepts and prototypes are obsolete before they are set in silicon.

The basic concept of the neuron has shifted from analog to digital, although many persist with analog.

On-chip learning has altered the concept of retraining.

In quick succession we've seen LSTM, attention, transformers, ViT, ...

This is a dizzying rate of change of basic functions which has outstripped the ability to timely manufacture the functions in silicon.

And then, of course, there is Chat GPT ...

While these ideas can be implemented in software, there's many a slip between CPU and SoC.

Akida itself has gone from single bit to 4-bit, to 8-bit compatibility. It flirted with LSTM but now has ViT and the proprietary TeNNs, which we are told is even better than the basic digital SNN implementation of Akida 1000 with its secret sauce, which is still at the leading edge of COTS SOTA.

Akida is not some mere appendage to the AGI revolution - it is the optimal gateway between the real-world and the cyberverse.

Akida is surfing the tidalwave of the technologically disruptive earthquake.

So yes, $2.34 may have been overpriced based on income - it is insignificant when based on potential.


Note: This is not investment advice - it is the distillation of shiraz and hot crossed buns (with lashings of butter).
Dio I agree the valuation is limitless if things go the right direction and im not an expert to give a value of BRN my value priced with now forward looking statement.

The interesting part is none of the holders Peter, Anil, Dimitro and the Osserieran's never sold shares so you need to look are why not clearly if BRN was a pumped hype stock these significant holders would have cashed in and moved on they been around longer then a year at that point lol. They did not sell why well they must value the shares higher then that or see future value north of that.

I know some wl say Peter sold share what he sold is insignificant to his holding and some if the sale was a donation.

You are also correct how fast things progress it's nuts.

My hope is we gain strong traction in the LLM sector and at that point current revenue will be second thought when the market sees that potential.
 
  • Like
  • Fire
Reactions: 19 users
Just on LLM.

Happy Easter as well to those that celebrate it.

Appears Microsoft researching 1-bit LLM.

Will this be of benefit to us and our 1-bit edge learning layer where inputs and weights are 1-bit...or am I on the wrong thinking. @Diogenese

In the below they seem to think neuromorphic architectures would excel with it.



The Future of AI Efficiency: 1-Bit LLMs Explained​


Vasu Rao

Vasu Rao​

Executive Product Management Leader Specialized…​

Published Mar 26, 2024
+ Follow
Have you ever wondered how much energy training a powerful language model takes? The answer might surprise you. A single training run can gulp down an astounding five megawatt-hours of electricity, equivalent to the annual consumption of several American households. As AI continues to evolve, this energy footprint becomes a pressing concern. The hefty energy demands of training LLMs strain budgets and resources. Cloud providers, research institutions, and even society feel the impact. 1-bit LLMs, with their dramatic efficiency gains, offer a path toward lower costs and a greener future for AI.
The world of large language models (LLMs) constantly evolves, pushing the boundaries of what AI can achieve. Enter 1-bit LLMs, a groundbreaking innovation from Microsoft that promises a significant leap forward in efficiency and accessibility. But what challenges do 1-bit LLMs aim to solve, and why is it a game-changer?
The Challenge: LLM Gluttony
Despite their impressive capabilities, current LLMs have a significant drawback: they are resource-hungry beasts. Training and running these models require massive computational power and electricity. Most commonly, contemporary LLMs from various players like OpenAI (GPT-3), Google (LaMDA, PaLM, Gemini), Meta (LLaMA), and Anthropic (Claude) utilize 32-bit floating-point precision for their parameters. This high precision allows for complex calculations and nuanced representations within the model, but it comes at a cost – immense computational resources.
Microsoft's Ingenious Solution: The 1-Bit LLM
Microsoft researchers introduced the concept of 1-bit LLMs, a novel architecture that utilizes a single binary digit (0 or 1) for each parameter within the model. This minor change dramatically reduces the memory footprint and computational requirements compared to traditional LLMs.
Why 1-Bit LLMs Matter
The efficiency gains of 1-bit LLMs open doors to exciting possibilities:

  • Democratization of AI: By lowering the resource barrier, 1-bit LLMs make AI technology more accessible to smaller companies and researchers who may not have access to robust computing infrastructure.
  • Wider deployment: The reduced footprint allows deployment of on-edge devices with limited resources, paving the way for on-device AI applications.
  • Increased scalability: The efficiency gains enable training even larger and more powerful LLMs without encountering insurmountable resource constraints.

Technical Deep Dive
Quantization Techniques:
Large Language Models (LLMs) traditionally rely on high-precision numbers (often 32-bit floating-point) to represent the vast amount of information they learn. Quantization is a technique for reducing the number of bits used for these parameters, leading to a smaller model footprint and lower computational demands. 1-bit LLMs represent the most extreme form of quantization, using a single bit (0 or 1) for each parameter. This significantly reduces the model size and computational needs compared to conventional LLMs.
Training Challenges:
Training 1-bit LLMs presents unique challenges compared to traditional models. One hurdle is the need for specialized training algorithms that can effectively learn with such limited precision. Existing training algorithms designed for high-precision models may not translate well to the binary world of 1-bit LLMs. Additionally, achieving convergence during training can be more difficult due to the limited representational capabilities of 1-bit parameters. Researchers are actively developing new training methods to address these challenges and unlock the full potential of 1-bit LLMs.
Comparison with Recent Work:
Microsoft recently introduced a significant advancement in 1-bit LLM research with BitNet b1.58. This variant utilizes a ternary system, assigning values of -1, 0, or 1 to each parameter. This offers a slight increase in representational power compared to the pure binary system of traditional 1-bit LLMs. Interestingly, BitNet b1.58 achieves performance on par with full-precision models while maintaining significant efficiency gains in terms of memory footprint and computational requirements. This development highlights the ongoing research efforts and promising future of 1-bit LLM technology.
Beyond Efficiency: Use Cases and Algorithm Advancements
The benefits of 1-bit LLMs extend beyond just resource savings. They can potentially:

  • Boost performance in specific tasks: The 1-bit representation's inherent simplicity might improve performance in applications like text classification or sentiment analysis.
  • Drive advancements in hardware design: The unique requirements of 1-bit LLMs could inspire the development of specialized hardware architectures optimized for their efficient operation.

Further Exploration: Hardware Advancements on the Horizon
The unique, binary nature of 1-bit LLMs could inspire the development of specialized hardware architectures beyond traditional CPUs and GPUs. Here are some potential areas of exploration for major chipmakers:

  • In-Memory Computing: Companies like Intel, with its "Xeon with Optane DC Persistent Memory," and Samsung, with its "Processing-in-Memory" (PIM) solutions, are exploring architectures that move computations closer to the memory where data resides. This could prove highly beneficial for 1-bit LLMs, as frequent memory access for parameter updates is crucial. The goal: significantly reduce latency and improve overall processing efficiency.
  • Neuromorphic Computing: Inspired by the human brain, neuromorphic chips attempt to mimic the structure and function of biological neurons. Companies like IBM with their TrueNorth and Cerebras Systems with their Wafer-Scale Engine are leaders in this field. Neuromorphic architectures could excel at the low-precision, binary operations that 1-bit LLMs rely on. The goal: achieve ultra-low power consumption while maintaining high performance for specific AI tasks.
  • Specialized Logic Units (SLUs): These custom-designed circuits could be tailored to handle the mathematical operations of 1-bit LLM training and inference. Companies like Google with their Tensor Processing Units (TPUs) and Nvidia with their Tensor Cores have experience in this area. The goal is to achieve significant performance gains and lower power consumption than general-purpose CPUs or GPUs for 1-bit LLM tasks.

These potential hardware advancements and ongoing research in 1-bit LLM algorithms hold promise for creating a new generation of efficient and powerful AI models.
Weighing the Pros and Cons
While 1-bit LLMs offer compelling advantages, there are potential drawbacks to consider:

  • Potential accuracy trade-offs: Depending on the specific task, using a single bit might lead to a slight decrease in accuracy compared to higher-precision models.
  • New research is needed: Optimizing training algorithms and techniques for 1-bit LLMs is an ongoing area of study.

Limitations and the Road Ahead
1-bit LLMs are still in their initial stages of development, and there are limitations to address:

  • Task-specific optimization: Identifying the tasks and applications where 1-bit LLMs excel requires further research.
  • Fine-tuning techniques: Developing effective methods for fine-tuning 1-bit LLMs for specific tasks is crucial for achieving optimal performance.

The Future of 1-Bit LLMs
The emergence of 1-bit LLMs signifies a significant step towards more efficient and accessible AI. While challenges remain, the potential for broader deployment, lower resource consumption, and even performance improvements in specific tasks make 1-bit LLMs a technology worth watching closely. As research progresses, we can expect 1-bit LLMs to play a transformative role in democratizing AI and unlocking their full potential.
Educational Resources:


Akida layers​

The sections below list the available layers for Akida 1.0 and Akida 2.0. Those layers are obtained from converting a quantized model to Akida and are thus automatically defined during conversion. Akida layers only perform integer operations using 8-bit or 4-bit quantized inputs and weights. The exception is FullyConnected layers performing edge learning, where both inputs and weights are 1-bit.
 
  • Like
  • Love
  • Fire
Reactions: 28 users

Diogenese

Top 20
Just on LLM.

Happy Easter as well to those that celebrate it.

Appears Microsoft researching 1-bit LLM.

Will this be of benefit to us and our 1-bit edge learning layer where inputs and weights are 1-bit...or am I on the wrong thinking. @Diogenese

In the below they seem to think neuromorphic architectures would excel with it.



The Future of AI Efficiency: 1-Bit LLMs Explained​


Vasu Rao

Vasu Rao​

Executive Product Management Leader Specialized…​

Published Mar 26, 2024
+ Follow
Have you ever wondered how much energy training a powerful language model takes? The answer might surprise you. A single training run can gulp down an astounding five megawatt-hours of electricity, equivalent to the annual consumption of several American households. As AI continues to evolve, this energy footprint becomes a pressing concern. The hefty energy demands of training LLMs strain budgets and resources. Cloud providers, research institutions, and even society feel the impact. 1-bit LLMs, with their dramatic efficiency gains, offer a path toward lower costs and a greener future for AI.
The world of large language models (LLMs) constantly evolves, pushing the boundaries of what AI can achieve. Enter 1-bit LLMs, a groundbreaking innovation from Microsoft that promises a significant leap forward in efficiency and accessibility. But what challenges do 1-bit LLMs aim to solve, and why is it a game-changer?
The Challenge: LLM Gluttony
Despite their impressive capabilities, current LLMs have a significant drawback: they are resource-hungry beasts. Training and running these models require massive computational power and electricity. Most commonly, contemporary LLMs from various players like OpenAI (GPT-3), Google (LaMDA, PaLM, Gemini), Meta (LLaMA), and Anthropic (Claude) utilize 32-bit floating-point precision for their parameters. This high precision allows for complex calculations and nuanced representations within the model, but it comes at a cost – immense computational resources.
Microsoft's Ingenious Solution: The 1-Bit LLM
Microsoft researchers introduced the concept of 1-bit LLMs, a novel architecture that utilizes a single binary digit (0 or 1) for each parameter within the model. This minor change dramatically reduces the memory footprint and computational requirements compared to traditional LLMs.
Why 1-Bit LLMs Matter
The efficiency gains of 1-bit LLMs open doors to exciting possibilities:

  • Democratization of AI: By lowering the resource barrier, 1-bit LLMs make AI technology more accessible to smaller companies and researchers who may not have access to robust computing infrastructure.
  • Wider deployment: The reduced footprint allows deployment of on-edge devices with limited resources, paving the way for on-device AI applications.
  • Increased scalability: The efficiency gains enable training even larger and more powerful LLMs without encountering insurmountable resource constraints.

Technical Deep Dive
Quantization Techniques:
Large Language Models (LLMs) traditionally rely on high-precision numbers (often 32-bit floating-point) to represent the vast amount of information they learn. Quantization is a technique for reducing the number of bits used for these parameters, leading to a smaller model footprint and lower computational demands. 1-bit LLMs represent the most extreme form of quantization, using a single bit (0 or 1) for each parameter. This significantly reduces the model size and computational needs compared to conventional LLMs.
Training Challenges:
Training 1-bit LLMs presents unique challenges compared to traditional models. One hurdle is the need for specialized training algorithms that can effectively learn with such limited precision. Existing training algorithms designed for high-precision models may not translate well to the binary world of 1-bit LLMs. Additionally, achieving convergence during training can be more difficult due to the limited representational capabilities of 1-bit parameters. Researchers are actively developing new training methods to address these challenges and unlock the full potential of 1-bit LLMs.
Comparison with Recent Work:
Microsoft recently introduced a significant advancement in 1-bit LLM research with BitNet b1.58. This variant utilizes a ternary system, assigning values of -1, 0, or 1 to each parameter. This offers a slight increase in representational power compared to the pure binary system of traditional 1-bit LLMs. Interestingly, BitNet b1.58 achieves performance on par with full-precision models while maintaining significant efficiency gains in terms of memory footprint and computational requirements. This development highlights the ongoing research efforts and promising future of 1-bit LLM technology.
Beyond Efficiency: Use Cases and Algorithm Advancements
The benefits of 1-bit LLMs extend beyond just resource savings. They can potentially:

  • Boost performance in specific tasks: The 1-bit representation's inherent simplicity might improve performance in applications like text classification or sentiment analysis.
  • Drive advancements in hardware design: The unique requirements of 1-bit LLMs could inspire the development of specialized hardware architectures optimized for their efficient operation.

Further Exploration: Hardware Advancements on the Horizon
The unique, binary nature of 1-bit LLMs could inspire the development of specialized hardware architectures beyond traditional CPUs and GPUs. Here are some potential areas of exploration for major chipmakers:

  • In-Memory Computing: Companies like Intel, with its "Xeon with Optane DC Persistent Memory," and Samsung, with its "Processing-in-Memory" (PIM) solutions, are exploring architectures that move computations closer to the memory where data resides. This could prove highly beneficial for 1-bit LLMs, as frequent memory access for parameter updates is crucial. The goal: significantly reduce latency and improve overall processing efficiency.
  • Neuromorphic Computing: Inspired by the human brain, neuromorphic chips attempt to mimic the structure and function of biological neurons. Companies like IBM with their TrueNorth and Cerebras Systems with their Wafer-Scale Engine are leaders in this field. Neuromorphic architectures could excel at the low-precision, binary operations that 1-bit LLMs rely on. The goal: achieve ultra-low power consumption while maintaining high performance for specific AI tasks.
  • Specialized Logic Units (SLUs): These custom-designed circuits could be tailored to handle the mathematical operations of 1-bit LLM training and inference. Companies like Google with their Tensor Processing Units (TPUs) and Nvidia with their Tensor Cores have experience in this area. The goal is to achieve significant performance gains and lower power consumption than general-purpose CPUs or GPUs for 1-bit LLM tasks.

These potential hardware advancements and ongoing research in 1-bit LLM algorithms hold promise for creating a new generation of efficient and powerful AI models.
Weighing the Pros and Cons
While 1-bit LLMs offer compelling advantages, there are potential drawbacks to consider:

  • Potential accuracy trade-offs: Depending on the specific task, using a single bit might lead to a slight decrease in accuracy compared to higher-precision models.
  • New research is needed: Optimizing training algorithms and techniques for 1-bit LLMs is an ongoing area of study.

Limitations and the Road Ahead
1-bit LLMs are still in their initial stages of development, and there are limitations to address:

  • Task-specific optimization: Identifying the tasks and applications where 1-bit LLMs excel requires further research.
  • Fine-tuning techniques: Developing effective methods for fine-tuning 1-bit LLMs for specific tasks is crucial for achieving optimal performance.

The Future of 1-Bit LLMs
The emergence of 1-bit LLMs signifies a significant step towards more efficient and accessible AI. While challenges remain, the potential for broader deployment, lower resource consumption, and even performance improvements in specific tasks make 1-bit LLMs a technology worth watching closely. As research progresses, we can expect 1-bit LLMs to play a transformative role in democratizing AI and unlocking their full potential.
Educational Resources:


Akida layers​

The sections below list the available layers for Akida 1.0 and Akida 2.0. Those layers are obtained from converting a quantized model to Akida and are thus automatically defined during conversion. Akida layers only perform integer operations using 8-bit or 4-bit quantized inputs and weights. The exception is FullyConnected layers performing edge learning, where both inputs and weights are 1-bit.
... and the wheel turns ...

I'll bet PvdM is having a quite chuckle.
 
  • Like
  • Haha
  • Fire
Reactions: 21 users
  • Fire
  • Like
  • Haha
Reactions: 9 users

Kachoo

Regular
... and the wheel turns ...

I'll bet PvdM is having a quite chuckle.
Did PvdM not say something like this lol

Just for reference.



Why research this you can just order 1 bit now!
 
  • Like
  • Fire
Reactions: 8 users

Boab

I wish I could paint like Vincent
In the last few years, the AI field has seen the most rapid advancement in technology ever.
Concepts and prototypes are obsolete before they are set in silicon.

The basic concept of the neuron has shifted from analog to digital, although many persist with analog.

On-chip learning has altered the concept of retraining.

In quick succession we've seen LSTM, attention, transformers, ViT, ...

This is a dizzying rate of change of basic functions which has outstripped the ability to timely manufacture the functions in silicon.

And then, of course, there is Chat GPT ...

While these ideas can be implemented in software, there's many a slip between CPU and SoC.

Akida itself has gone from single bit to 4-bit, to 8-bit compatibility. It flirted with LSTM but now has ViT and the proprietary TeNNs, which we are told is even better than the basic digital SNN implementation of Akida 1000 with its secret sauce, which is still at the leading edge of COTS SOTA.

Akida is not some mere appendage to the AGI revolution - it is the optimal gateway between the real-world and the cyberverse.

Akida is surfing the tidalwave of the technologically disruptive earthquake.

So yes, $2.34 may have been overpriced based on income - it is insignificant when based on potential.


Note: This is not investment advice - it is the distillation of shiraz and hot crossed buns (with lashings of butter).
Before I got to the last sentence I thought how many reds has this guy had??😁😁
Love the enthusiasm along with the large brain.
Nice one Dodgy, enjoy.
 
  • Like
  • Haha
  • Fire
Reactions: 9 users
Happy Easter

1711827806802.gif
 
  • Like
  • Love
  • Fire
Reactions: 7 users

BrainShit

Regular
Can I ask you please about the Arm Cortex M55 and if we know if BRN has or can be integrated into this yet ?.
Yes, it can.... just take a look at the following graphic

Screenshot_20230521_120933_YouTube.jpg
 
  • Like
  • Love
  • Fire
Reactions: 27 users

TopCat

Regular
  • Like
  • Fire
  • Love
Reactions: 35 users

Tothemoon24

Top 20
Most likely posted from early March .
Impressive 🐰


Akida Edge AI Box: Edge Computing with Neuromorphic Technology​

author avatar

Jessica Miley


https://www.wevolver.com/profile/brainchip
Akida Edge AI Box: Edge Computing with Neuromorphic Technology


BrainChip Holdings Ltd. Initiates Pre-Orders for the Akida Edge AI Box, Elevating Edge AI Capabilities through a Strategic Alliance with VVDN Technologies​

Edge AI
- Edge Processors
- Neuromorphic Computing
BrainChip Holdings Ltd. has recently announced that it has opened pre-orders for the Akida Edge AI Box, a significant milestone in the field of Edge AI. Through collaboration with VVDN Technologies, BrainChip has made significant progress in developing high-performance and energy-efficient artificial intelligence solutions that can be applied to various industries and use cases.

Unveiling the Power of the Akida Edge AI Box​

The Akida Edge AI Box is engineered to harness the potential of neuromorphic computing, an innovative approach that mimics the human brain's neural networks. This technology enables the device to perform complex cognitive tasks more efficiently, using significantly less power than traditional computing methods.
Designed for versatility, the Akida Edge AI Box finds its application across various sectors, including retail, security, smart cities, automotive, and industrial processes. It aims to offer a solution that not only reduces operational costs but also enhances data privacy and security by processing data on the device, thereby minimising the need for constant cloud connectivity.

Key Features and Applications​

  • Ultra-Low Power Consumption: The device operates on minimal power, making it ideal for deployment in energy-sensitive environments.
  • High-Performance AI Capabilities: With the ability to handle high-level computations and process vast amounts of data, the Akida Edge AI Box is perfect for complex applications like image processing and real-time analytics.
  • Adaptability and Learning: Mimicking the plasticity of the human brain, the device can learn from new information and adjust its processing accordingly, enhancing its efficiency over time.
  • Real-Time Processing: The Akida Edge AI Box excels in scenarios requiring immediate data analysis and decision-making, such as autonomous vehicles and smart surveillance systems.
  • Robust and Resilient: Engineered to be fault-tolerant, the device ensures reliable operation even in challenging environmental conditions.

Technical Specifications​

eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNzA5NTUwNjQ2NTgyLVNjcmVlbnNob3QgMjAyNC0wMy0wNCAxMjEwMTUucG5nIiwiZWRpdHMiOnsicmVzaXplIjp7IndpZHRoIjo5NTAsImZpdCI6ImNvdmVyIn19fQ==

Priced at $799, the Akida Edge AI Box not only stands out for its technological prowess but also for its affordability, making advanced neuromorphic computing accessible to a broader range of industries.
eyJidWNrZXQiOiJ3ZXZvbHZlci1wcm9qZWN0LWltYWdlcyIsImtleSI6ImZyb2FsYS8xNzA5NTUwNDI5NjI5LWJyYWluYm94LmpwZyIsImVkaXRzIjp7InJlc2l6ZSI6eyJ3aWR0aCI6OTUwLCJmaXQiOiJjb3ZlciJ9fX0=
The Akida is best in class compared to current market products. Image credit: BrainChip.

Bringing Edge AI Everywhere: Application examples​

Security and Surveillance​

the Akida Edge AI Box's capabilities have the potential to significantly transform the way monitoring and threat detection are carried out.. Its ability to process data locally, with ultra-low latency, ensures real-time analysis of video feeds, detecting anomalies, unauthorized access, or suspicious behaviors instantly. This rapid processing can significantly enhance security measures in both public and private spaces, offering a more robust response to potential threats.

Smart Factories​

For smart factories, the Akida Edge AI Box brings the potential for improved automation, predictive maintenance, and operational efficiency. By processing data on-site, it can monitor equipment health in real-time, predict failures before they occur, and optimize manufacturing processes. This not only reduces downtime but also extends the lifespan of machinery, leading to significant cost savings and increased production efficiency.

Smart Retail​

In smart retail environments, the Akida Edge AI Box can personalize the shopping experience through real-time analytics. It can manage inventory, optimize store layouts based on customer behavior analytics, and offer targeted promotions to customers. This level of personalization and efficiency can enhance customer satisfaction and loyalty, driving sales and improving retail operations.

Smart Cities​

The Akida Edge AI Box can play a pivotal role in developing smart city infrastructure, from traffic management and waste management to energy conservation. Its capacity for edge processing allows for the real-time analysis of data from various sensors across the city, enabling intelligent decision-making that can improve city services, reduce congestion, enhance public safety, and lower environmental impact.

Warehouses​

In warehouses, the Akida Edge AI Box can streamline operations by optimizing logistics, managing inventory in real-time, and automating the tracking and dispatching of goods. This not only improves operational efficiency but also reduces the likelihood of errors, ensuring that the right products are always in stock and delivered on time.
By bringing AI processing capabilities directly to the edge, the Akida Edge AI Box enables these sectors to operate more efficiently, safely, and sustainably. Its introduction represents a significant step forward in the deployment of intelligent, secure, and customized devices and services for multi-sensor environments, catering to the specific needs of these diverse applications.

The Future of Edge Computing​

The collaboration between BrainChip and VVDN Technologies in developing the Akida Edge AI Box is a testament to the power of strategic partnerships in driving technological innovation. This device is set to redefine the landscape of Edge computing, ushering in a new era of intelligent, efficient, and autonomous systems.
By addressing the critical challenges of energy consumption, processing power, and data privacy, the Akida Edge AI Box is poised to play a crucial role in the evolution of Edge AI. Its launch not only meets the current market demand for more efficient computing solutions but also lays the groundwork for the future of intelligent devices and systems across a multitude of sectors.
As industries continue to seek smarter, more efficient technologies, the Akida Edge AI Box stands as a beacon of progress in the realm of Edge computing, promising a future where artificial intelligence is both profoundly intelligent and sustainably efficient.
Order the Akida Edge AI Box here.
 
  • Like
  • Love
  • Fire
Reactions: 41 users

Esq.111

Fascinatingly Intuitive.
This podcast goes for an hour so might try and listen later. Happy Easter!




View attachment 60057
Good Morning TopCat & Fellow Chippers,

Great find TopCat.

The good Doctor Joseph Guerci knows a good thing when he sees it.

Time stamp 42:30 ish..

VERY INTERESTING.

Worth listening to the whole interview and particularly the above time mark,... right through to the end.

Regards,
Esq.
 
Last edited:
  • Like
  • Love
  • Fire
Reactions: 50 users

Boab

I wish I could paint like Vincent
This podcast goes for an hour so might try and listen later. Happy Easter!




View attachment 60057
Joe says "this is the revolution, this is the thing that changes everything" referring to SNN's and Brainchip. Intel and IBM get a small mention but Brainchip is the main character.
Wow, what a fabulous recommendation he gives us.
 
Last edited:
  • Like
  • Fire
  • Love
Reactions: 74 users

Iseki

Regular
Good Morning TopCat & Fellow Chippers,

Great find TopCat.

The good Doctor Joseph Guerci knows a good thing when he sees it.

Time stamp 42:30 ish..

VERY INTERESTING.

Worth listening to the whole interview and particularly the above time mark,... right through to the end.

Regards,
Esq.
Hi Esq,

thank you for your ever joyous posts.

Yes, great wrap for Brainchip.
Then at 48:00 mark, they start talking about the Valley of Death - what tech startups can do to survive the time before revenue.

For some analysis on this, check out this article
 
  • Like
  • Love
  • Fire
Reactions: 13 users
Good Morning TopCat & Fellow Chippers,

Great find TopCat.

The good Doctor Joseph Guerci knows a good thing when he sees it.

Time stamp 42:30 ish..

VERY INTERESTING.

Worth listening to the whole interview and particularly the above time mark,... right through to the end.

Regards,
Esq.
1711837964263.gif
 
  • Haha
  • Like
Reactions: 4 users

FJ-215

Regular
This podcast goes for an hour so might try and listen later. Happy Easter!




View attachment 60057
Good morning @TopCat

Imagine if these words had been spoken by a tech heavy weight like Tim Cook or Jensen Huang.

WOW!!
 
  • Like
  • Fire
  • Wow
Reactions: 17 users

hotty4040

Regular
Joe says "this is the revolution, this is the thing that changes everything" referring to SNN's and Brainchip. Intel and IBM get a small mention but Brainchip is the main character.
Wow, what a fabulous recommendation he gives us.


I liked this FTCN Episode 97, and I liked it a lot >>>>>> Akida Ballista indeed <<<<<

hotty...
 
  • Like
  • Fire
Reactions: 17 users

Bravo

If ARM was an arm, BRN would be its biceps💪!

Screenshot 2024-03-31 at 10.41.04 am.png

South Korea and EU to launch joint research on brain-inspired semiconductors​

Kicking off as early as July, the project is set to channel €12 million over three years
By Kwak Soo-keun,
Yeom Hyun-a
Published 28 March 2024



Lee Jong-ho, Minister of Science and ICT of Korea, speaking at the second ROK-EU Digital Partnership Council meeting held at the Bérémont Building in Brussels, Belgium, on Mar. 26, 2024./Yonhap News


Lee Jong-ho, Minister of Science and ICT of Korea, speaking at the second ROK-EU Digital Partnership Council meeting held at the Bérémont Building in Brussels, Belgium, on Mar. 26, 2024./Yonhap News

Starting in July, South Korea and the European Union (EU) are set to begin collaborative research on neuromorphic chips, an advanced type of semiconductor modeled after the human brain’s neural network.

This initiative involves an investment of €12 million (16.2 billion won) spread over three years. Additionally, by the year’s end, Korea and the EU plan to extend their cooperation to areas like 6G (sixth-generation mobile communications), with an allocated budget of €5 million (6.8 billion won) for the next three years.

The Ministry of Science and ICT (MSIT) in Korea shared on Mar. 27 that the collaborative research agenda was established during the ‘2nd Republic of Korea-EU Digital Partnership Council’ held in Brussels, Belgium. As per the agreement, the collaborative efforts between Korea and the EU will span various domains including semiconductors, 6G, AI, and cybersecurity.

The partnership is expected to widen next year when Korea acquires Associate Membership in Horizon Europe, the world’s most extensive research and innovation program led by the EU. Horizon Europe has a budget of €95.5 billion (130 trillion won) for the seven years from 2021 to 2027. Korea, having been granted associate membership on Mar. 25, marks the first Asian nation to join this program.

This association will open doors for Korean researchers to engage in international collaborative projects funded by Horizon Europe starting next year. As an associate member, Korea will contribute financially to joint research endeavors aimed at solving global challenges and bolstering industrial competitiveness.

 
Last edited:
  • Like
  • Love
  • Fire
Reactions: 21 users

Terroni2105

Founding Member
  • Like
  • Love
  • Fire
Reactions: 28 users

Kachoo

Regular
Brilliant ISL podcast @TopCat , thank you.


on another thing here is a post by Wevolver

Quite the pod cast Joe can't stress enough how powerful Akida is lol
 
  • Like
  • Fire
  • Love
Reactions: 28 users
Top Bottom