CHIPS
Regular
Where is the link to this article?
BRN Discussion Ongoing
- Thread starter TechGirl
- Start date
Where is the link to this article?
Food4 thought
Regular
For world class technology
And the power saving alone
They are pretty piss poor amounts for the stock if you ask me I know in in US dollars but still itās rubbish
FiveBucks
Regular
I'd be happy with those prices..... By the end of next year.
Where is the link to this article?
BRCHF - Brainchip Holdings Ltd Stock Price Forecast 2024, 2025, 2030 to 2050
Explore BRCHF - Brainchip Holdings Ltd Stock Price Forecast: Accurate short-term and long-term price predictions up to 2050
Tothemoon24
Top 20
Useful Sensors ; start up tech company from California .
I canāt find much info about them ,
Useful Sensor's technology enables the device to operate entirely offline with no need to send information to the cloud,
SAN FRANCISCO--(BUSINESS WIRE)--Useful Sensors, a leader in developing advanced AI for everyday objects, today announced Torre, a self-contained translation solution delivering superior language accuracy and speed on a single, dual-screened device. Functioning independent of cellular or internet connection while delivering industry-leading language accuracy and speed, this device sets a new standard for performance and privacy in cross-language communication.
Torre is the next evolution in this mission, as Useful Sensor's technology enables the device to operate entirely offline with no need to send information to the cloud, thus ensuring complete data privacy compared to smartphones, tablets, and other network-connected devices. This makes it an ideal choice for users concerned about data security, or those who just wish to keep their conversations private while utilizing transformative AI technology.
The device features an intuitive dual-screen design that displays translations in real time, enabling natural, fluid conversations. This versatile functionality is perfect for a wide range of settings, from travel and hospitality to medical consultations and legal services. With its user-friendly interface and robust performance, Torre is poised to become an essential tool for bridging communication gaps.
āJust like how subtitles help you follow along with the movie without missing a beat, our device displays real-time translations, allowing conversations to flow naturally. It's about creating an experience where language barriers simply fade away, enabling meaningful interactions without awkward pauses or misunderstandings.ā - Pete Warden, CEO and CoFounder.
Useful Sensors is actively engaging with potential partners across various industries, including healthcare, legal services, and consumer electronics. These discussions underscore the growing recognition of Torreās potential to revolutionize communication and enhance user experiences across diverse settings.
Torre is now available to customers on a limited basis. For more information, visit usefulsensors.com.
About Useful Sensors
Useful Sensors specializes in developing advanced AI and machine learning solutions for low-cost hardware. Focusing on privacy and user-friendly design, the company integrates innovative technology into everyday devices, making them more accessible and practical for a wide range of applications. Co-founded by Pete Warden, creator of TinyML, and Manjunath Kudlur, former lead at Nvidia and Cerebras, Useful Sensors is led by a team renowned for their expertise in AI. The company's recent work is setting standards in AI model performance besting the results of industry heavyweights. With a commitment to flawless execution and continuous innovation, Useful Sensors is at the forefront of creating smarter, more intuitive devices.
I canāt find much info about them ,
Useful Sensor's technology enables the device to operate entirely offline with no need to send information to the cloud,
Useful Sensors Unveils Torre, a Groundbreaking Real-Time Translation Device with Superior Privacy and Accuracy
October 01, 2024 09:00 AM Eastern Daylight TimeSAN FRANCISCO--(BUSINESS WIRE)--Useful Sensors, a leader in developing advanced AI for everyday objects, today announced Torre, a self-contained translation solution delivering superior language accuracy and speed on a single, dual-screened device. Functioning independent of cellular or internet connection while delivering industry-leading language accuracy and speed, this device sets a new standard for performance and privacy in cross-language communication.
Useful Sensors, co-founded by Pete Warden, creator of TinyML, and Manjunath Kudlur, a former lead at Nvidia and Cerebras, leverages years of expertise in AI and machine learning to develop innovative products that work seamlessly on low-cost hardware. The company's mission is to make interacting with machines, electronic devices, and everyday appliances as easy as communicating with a person while maintaining user privacy and data security.
Torre is the next evolution in this mission, as Useful Sensor's technology enables the device to operate entirely offline with no need to send information to the cloud, thus ensuring complete data privacy compared to smartphones, tablets, and other network-connected devices. This makes it an ideal choice for users concerned about data security, or those who just wish to keep their conversations private while utilizing transformative AI technology.
The device features an intuitive dual-screen design that displays translations in real time, enabling natural, fluid conversations. This versatile functionality is perfect for a wide range of settings, from travel and hospitality to medical consultations and legal services. With its user-friendly interface and robust performance, Torre is poised to become an essential tool for bridging communication gaps.
āJust like how subtitles help you follow along with the movie without missing a beat, our device displays real-time translations, allowing conversations to flow naturally. It's about creating an experience where language barriers simply fade away, enabling meaningful interactions without awkward pauses or misunderstandings.ā - Pete Warden, CEO and CoFounder.
Useful Sensors is actively engaging with potential partners across various industries, including healthcare, legal services, and consumer electronics. These discussions underscore the growing recognition of Torreās potential to revolutionize communication and enhance user experiences across diverse settings.
Torre is now available to customers on a limited basis. For more information, visit usefulsensors.com.
About Useful Sensors
Useful Sensors specializes in developing advanced AI and machine learning solutions for low-cost hardware. Focusing on privacy and user-friendly design, the company integrates innovative technology into everyday devices, making them more accessible and practical for a wide range of applications. Co-founded by Pete Warden, creator of TinyML, and Manjunath Kudlur, former lead at Nvidia and Cerebras, Useful Sensors is led by a team renowned for their expertise in AI. The company's recent work is setting standards in AI model performance besting the results of industry heavyweights. With a commitment to flawless execution and continuous innovation, Useful Sensors is at the forefront of creating smarter, more intuitive devices.
Last edited:
NVIDIA, Qualcomm, Google, Samsung team up to take on Intel, AMD with new CPU architecture
Although it might be some time before RISC-V chips appear in mainstream PCs and servers, the architecture is already making waves in the AI and automotive sectors
Great news to hear that Ken Wu is presenting at the tinyML Foundation in Taipei, Taiwan...I thought that he had left
his role as Senior Design Engineer at Brainchip, but no....yet another 6 year staffer, highly intelligent, whom I'm glad is
backing AKIDA....we have the team....let's go Brainchip
his role as Senior Design Engineer at Brainchip, but no....yet another 6 year staffer, highly intelligent, whom I'm glad is
backing AKIDA....we have the team....let's go Brainchip
Great news to hear that Ken Wu is presenting at the tinyML Foundation in Taipei, Taiwan...I thought that he had left
his role as Senior Design Engineer at Brainchip, but no....yet another 6 year staffer, highly intelligent, whom I'm glad is
backing AKIDA....we have the team....let's go Brainchip
Amazing!
Pom down under
Top 20
Tothemoon24
Top 20
SCOTTSDALE, Ariz.--(BUSINESS WIRE)--onsemi (Nasdaq: ON) announced today its wafer fab in East Fishkill (EFK), New York, has been accredited by the Defense Microelectronics Activity (DMEA) of the U.S. Department of Defense (DoD) as a Category 1A Trusted Supplier. The accreditation enables onsemi to manufacture secure advanced semiconductors for a variety of critical aerospace and defense applications at the site. onsemi's East Fishkill fab is the only 300 mm power discrete and image sensor fab in the U.S. and the companyās second accredited fab in addition to its Gresham, Oregon site.
āPower and sensing technologies are vital in modern defense systems, providing the reliable energy and data required for robust, intelligent and efficient operations,ā said Sudhir Gopalswamy, group president, Intelligent Sensing and Analog and Mixed-Signal Group of onsemi. āWith this accreditation, onsemi is now positioned to offer our industry-leading power and sensing technologies through our secure and trusted manufacturing facilities that will enable the DoD to have the reliable and cutting-edge solutions they need.ā
Achieving the trusted foundry accreditation is a rigorous process that ensures a supplier meets comprehensive security and operational criteria, including the protection of mission-critical functions for trusted systems and networks. This thorough approach safeguards the integrity and confidentiality of onsemiās industry-leading intelligent power and sensing technologies. As a trusted supplier, onsemi is able to provide custom-designed, custom-manufactured or tailored power and sensing solutions for specific DoD applications.
Bravo
If ARM was an arm, BRN would be its bicepsšŖ!
Whoa!
ByColin Wood
October 29, 2024
(Getty Images)
The University of Texas at San Antonio on Tuesday announced itās received $4 million from the National Science Foundation to establish a specialized computing system that researchers said will be larger than any of its kind in the United States.
Researchers from the universityās Matrix AI Consortium for Human Well-Being will use the funding to develop The Neuromorphic Commons, or THOR, project. Neuromorphic computing is an approach to designing computers that mimic or take inspiration from the human brainās capacity to solve complex problems on the fly using relatively little energy.
Dhireesha Kudithipudi, an electrical engineering and computer science professor at the university whoās serving as the projectās principal investigator, told EdScoop the goal of the THOR project is to provide broad access to large-scale neuromorphic systems to U.S. researchers from a variety of disciplines, including computational neuroscience, life sciences, artificial intelligence, machine learning and physics.
āThe idea is that by providing access to an infrastructure like that, we hope that it enables researchers to have a richer understanding of the computational models on neuromorphic or neuro-inspired machine-learning or AI algorithms,ā Kudithipudi said. āBetter understanding of neuromorphic hardware also provides them a framework to benchmark their models across different platforms.ā
Kudithipudi, whose work has centered on energy-efficient computing, and who founded the Matrix lab, explained that neuromorphic systems are well-suited for applications that require rapid reaction times and when hardware is constrained by size, weight or power supply. She pointed out that the human brain ā easily the most sophisticated object in the known universe ā runs on about 20 watts of power, just enough to power a lightbulb.
āA lot of experiments in this domain have been limited to smaller datasets or smaller model sizes,ā she said. āBy providing access to this large infrastructure, people can look at scale how their models or how their systems can work or what energy benefit they can [get] compared to the machine-learning systems.ā
She said many prominent computer scientists and engineers have gestured toward the human brain as the next frontier for computing. Among these was John von Neumann, a mathematician and scientist who developed many of the foundational ideas used in game theory and modern computing. Kudithipudi said the THOR projectās collaborative nature takes inspiration from his legacy.
āWe started looking at a lot of the literature, going back to the 1800s, and ā¦ von Neumann himself, whoās considered the father of computing, heās an amazing collaborator. He knew how to bring people together and work across disciplines. In his last memoir he said if you really want to build these robust systems, you want to look at [the] brain as a source of inspiration. And this [idea] has come from so many researchers, across decades.ā
edscoop.com
Neuromorphic computing hub at University of Texas at San Antonio to be largest in U.S.
The Neuromorphic Commons, or THOR, project, intends to provide researchers from many disciplines better access to computing architectures that take inspiration from the human brain.ByColin Wood
October 29, 2024
The University of Texas at San Antonio on Tuesday announced itās received $4 million from the National Science Foundation to establish a specialized computing system that researchers said will be larger than any of its kind in the United States.
Researchers from the universityās Matrix AI Consortium for Human Well-Being will use the funding to develop The Neuromorphic Commons, or THOR, project. Neuromorphic computing is an approach to designing computers that mimic or take inspiration from the human brainās capacity to solve complex problems on the fly using relatively little energy.
Dhireesha Kudithipudi, an electrical engineering and computer science professor at the university whoās serving as the projectās principal investigator, told EdScoop the goal of the THOR project is to provide broad access to large-scale neuromorphic systems to U.S. researchers from a variety of disciplines, including computational neuroscience, life sciences, artificial intelligence, machine learning and physics.
āThe idea is that by providing access to an infrastructure like that, we hope that it enables researchers to have a richer understanding of the computational models on neuromorphic or neuro-inspired machine-learning or AI algorithms,ā Kudithipudi said. āBetter understanding of neuromorphic hardware also provides them a framework to benchmark their models across different platforms.ā
Kudithipudi, whose work has centered on energy-efficient computing, and who founded the Matrix lab, explained that neuromorphic systems are well-suited for applications that require rapid reaction times and when hardware is constrained by size, weight or power supply. She pointed out that the human brain ā easily the most sophisticated object in the known universe ā runs on about 20 watts of power, just enough to power a lightbulb.
āA lot of experiments in this domain have been limited to smaller datasets or smaller model sizes,ā she said. āBy providing access to this large infrastructure, people can look at scale how their models or how their systems can work or what energy benefit they can [get] compared to the machine-learning systems.ā
She said many prominent computer scientists and engineers have gestured toward the human brain as the next frontier for computing. Among these was John von Neumann, a mathematician and scientist who developed many of the foundational ideas used in game theory and modern computing. Kudithipudi said the THOR projectās collaborative nature takes inspiration from his legacy.
āWe started looking at a lot of the literature, going back to the 1800s, and ā¦ von Neumann himself, whoās considered the father of computing, heās an amazing collaborator. He knew how to bring people together and work across disciplines. In his last memoir he said if you really want to build these robust systems, you want to look at [the] brain as a source of inspiration. And this [idea] has come from so many researchers, across decades.ā
Neuromorphic computing hub at University of Texas at San Antonio to be largest in U.S. | EdScoop
The Neuromorphic Commons, or THOR, project, intends to provide researchers from many disciplines better access to computing architectures that take inspiration from the human brain.
edscoop.com
Bravo
If ARM was an arm, BRN would be its bicepsšŖ!
Some future competition...
Oct 29, 2024 04:06 PM IST
Artificial Intelligence (AI) is set to create one of the largest market opportunities in history, with estimates placing its potential value between $3.5 and 5.8 trillion. Capturing a significant slice of this market could redefine national economies, acting as a powerful engine of growth for decades to come. For India, harnessing AI is key to achieving the vision of a Viksit Bharat by 2047.
While AI has long been a subject of fascination, it has also seen cycles of breakthroughs and disappointments. A closer look reveals a critical flaw ā these breakthroughs come with enormous energy demands and costly, time-consuming training processes. If nothing changes, projections suggest that AIās power needs could surpass global energy production by 2035, with profound economic and environmental consequences. This demands a leap in computing hardware that could be dramatically more energy-efficient than what we have today.
Why is this leap necessary? It comes down to the ageing von Neumann architecture, the blueprint for all computers over the last 60 years. In this model, computation and memory are separated, which slows down operations and guzzles energy. For tasks requiring billions of calculations per second, like those used in AI, the von Neumann design has become a major bottleneck. Whatās worse, the data we generate and use in AI systems is often stored by large corporations, raising privacy concerns.
he solution might be closer than we thinkāinside our own heads. The human brain, weighing less than two kg and consuming just 20 watts of energy, is capable of performing billions of operations per second, all while seamlessly storing and processing information in the same place. This extraordinary efficiency has inspired a new approach to computing, one modelled on the brainās neural networks.
The concept of brain-inspired computing isnāt new. In the 1980s, visionary American engineer Carver Mead laid the groundwork for what could become the future of computing. Fast forward to the 2010s ā industry giants like Intel and IBM reignited interest in brain-like computing. With advanced fabrication technologies at their disposal, these companies attempted to mimic the brainās learning processes using traditional binary transistors and software-driven systems. Unsurprisingly, this brute-force approach failed.
The lesson was clear: To come anywhere close to the brainās computational efficiency, we need to reimagine computing with new circuit elements that can learn and adapt like biological neurons and synapses. We also needed to rethink the entire computing architecture, moving beyond the limitations of von Neumann systems where memory and processing are separated.
The race to develop brain-inspired computers is not just about mimicking the brainās processing power ā itās about doing so with the same energy efficiency and compactness that make the brain so remarkable. The question is: can we build machines that are as smart and efficient as the human brain? The challenge lies in creating computing systems that can store information in thousands of states and operate at the edge of chaos, just like the brain.
In a groundbreaking study published in Nature, a team led by Dr Sreetosh Goswami from the Indian Institute of Science, Bengaluru, invented a revolutionary molecular neuromorphic platform capable of storing and processing data in an astonishing 16,500 states ā leaving traditional transistor-based computers, which operate in just two states, far behind. By harnessing the dancing of ions within a molecular film, the team created a system that mimics the brainās intricate method of data processing. The molecules and ions, wiggling within the film, generate a multitude of unique memory states. Each movement was mapped to a distinct electrical signal ā essentially a computer that captures thousands of computing states excelling in both energy efficiency and space-saving potential.
The breakthrough doesnāt stop there. In a stunning technological leap, the team used their molecular platform to recreate NASAās iconic Pillars of Creation image from the James Webb Telescope on a simple tabletop setup. Whatās more, they achieved this feat 4,000 times faster and with 460 times less energy than a traditional computer would require.
With 14-bit precision, equivalent to 16,384 analog levels, this chip could transform fields ranging from Artificial Intelligence (AI) to scientific computing. Imagine training complex AI models, such as Large Language Models (LLMs), directly on personal devices like laptops and smartphonesāa process that currently relies on vast server farms and invasive personal data collection by big corporations. This invention could bring AI processing to individual users, offering unprecedented data privacy and democratising access to advanced AI tools. This is arguably one of the most disruptive computing innovations to emerge from India, with the potential to position the nation at the vanguard of global technological advancements.
This article is authored by Brainerd Prince, director, Centre for Thinking, Language and Communication, Plaksha University.
Indiaās step into neuromorphic computing and the AI revolution
By Brainerd PrinceOct 29, 2024 04:06 PM IST
Artificial Intelligence (AI) is set to create one of the largest market opportunities in history, with estimates placing its potential value between $3.5 and 5.8 trillion. Capturing a significant slice of this market could redefine national economies, acting as a powerful engine of growth for decades to come. For India, harnessing AI is key to achieving the vision of a Viksit Bharat by 2047.
While AI has long been a subject of fascination, it has also seen cycles of breakthroughs and disappointments. A closer look reveals a critical flaw ā these breakthroughs come with enormous energy demands and costly, time-consuming training processes. If nothing changes, projections suggest that AIās power needs could surpass global energy production by 2035, with profound economic and environmental consequences. This demands a leap in computing hardware that could be dramatically more energy-efficient than what we have today.
Why is this leap necessary? It comes down to the ageing von Neumann architecture, the blueprint for all computers over the last 60 years. In this model, computation and memory are separated, which slows down operations and guzzles energy. For tasks requiring billions of calculations per second, like those used in AI, the von Neumann design has become a major bottleneck. Whatās worse, the data we generate and use in AI systems is often stored by large corporations, raising privacy concerns.
he solution might be closer than we thinkāinside our own heads. The human brain, weighing less than two kg and consuming just 20 watts of energy, is capable of performing billions of operations per second, all while seamlessly storing and processing information in the same place. This extraordinary efficiency has inspired a new approach to computing, one modelled on the brainās neural networks.
The concept of brain-inspired computing isnāt new. In the 1980s, visionary American engineer Carver Mead laid the groundwork for what could become the future of computing. Fast forward to the 2010s ā industry giants like Intel and IBM reignited interest in brain-like computing. With advanced fabrication technologies at their disposal, these companies attempted to mimic the brainās learning processes using traditional binary transistors and software-driven systems. Unsurprisingly, this brute-force approach failed.
The lesson was clear: To come anywhere close to the brainās computational efficiency, we need to reimagine computing with new circuit elements that can learn and adapt like biological neurons and synapses. We also needed to rethink the entire computing architecture, moving beyond the limitations of von Neumann systems where memory and processing are separated.
The race to develop brain-inspired computers is not just about mimicking the brainās processing power ā itās about doing so with the same energy efficiency and compactness that make the brain so remarkable. The question is: can we build machines that are as smart and efficient as the human brain? The challenge lies in creating computing systems that can store information in thousands of states and operate at the edge of chaos, just like the brain.
In a groundbreaking study published in Nature, a team led by Dr Sreetosh Goswami from the Indian Institute of Science, Bengaluru, invented a revolutionary molecular neuromorphic platform capable of storing and processing data in an astonishing 16,500 states ā leaving traditional transistor-based computers, which operate in just two states, far behind. By harnessing the dancing of ions within a molecular film, the team created a system that mimics the brainās intricate method of data processing. The molecules and ions, wiggling within the film, generate a multitude of unique memory states. Each movement was mapped to a distinct electrical signal ā essentially a computer that captures thousands of computing states excelling in both energy efficiency and space-saving potential.
The breakthrough doesnāt stop there. In a stunning technological leap, the team used their molecular platform to recreate NASAās iconic Pillars of Creation image from the James Webb Telescope on a simple tabletop setup. Whatās more, they achieved this feat 4,000 times faster and with 460 times less energy than a traditional computer would require.
With 14-bit precision, equivalent to 16,384 analog levels, this chip could transform fields ranging from Artificial Intelligence (AI) to scientific computing. Imagine training complex AI models, such as Large Language Models (LLMs), directly on personal devices like laptops and smartphonesāa process that currently relies on vast server farms and invasive personal data collection by big corporations. This invention could bring AI processing to individual users, offering unprecedented data privacy and democratising access to advanced AI tools. This is arguably one of the most disruptive computing innovations to emerge from India, with the potential to position the nation at the vanguard of global technological advancements.
This article is authored by Brainerd Prince, director, Centre for Thinking, Language and Communication, Plaksha University.
Bravo
If ARM was an arm, BRN would be its bicepsšŖ!
Can anyone remember if MegaChips said BrainChip's technology is complimentary to Quadric's or are we in direct competition with them?
Oct 29, 2024, 12:41 ET
ā Collaboration Accelerates Joint Development of Semiconductor IP Suitable for Automotive Applications ā
KARIYA, Japan, Oct. 29, 2024 /PRNewswire/ -- DENSO CORPORATION and Quadric.inc have signed a development license agreement for a Neural Processing Unit (NPU) * 1, which is a semiconductor specialized for the arithmetic processing of AI. Through the agreement, DENSO will acquire the IP core license for Quadric's Chimera GPNPU * 2, and the two companies will co-develop intellectual property (NPU) for an in-vehicle semiconductor.
With the advancement of intelligent systems, such as automated driving and connected technologies, vehicles need to process in real time a vast amount of information obtained from on-board sensors, as well as inter-vehicle and cloud communications. For this reason, the performance of the in-vehicle SoC* 3, which can process large amounts of information at high speed and with efficiency, has become increasingly important.
DENSO and Quadric have been studying semiconductor IP (NPU) development using Chimera GPNPU (General Purpose Neural Processing Unit) for in-vehicle SoCs and have now decided to jointly develop an in-vehicle semiconductor IP (NPU) combining DENSO 's RISC-V* 4 based processor and Quadric's Chimera GPNPU.
DENSO 's RISC-V-based processor IP is compliant with ISO26262 ASIL * 5 D and is ideal for automotive applications, where safety is critical. As the intelligence of vehicles advances, AI will become a pivotal factor in determining the value of vehicles. Quadric's Chimera GPNPU has a unique architecture that can handle matrix and vector operations and scalar (control) code in one execution pipeline. This flexibility in handling diverse computational tasks makes it applicable to various AI functions. In addition, by using Chimera GPNPU, automotive system developers can add proprietary AI features to their products throughout the future.
By pursuing joint development of a new semiconductor IP (NPU), DENSO and Quadric aim to contribute to the realization of in-vehicle SoCs that can flexibly respond to changes in AI trends over a long period of time, from product development to post-release, and to promote the development of highly secure AD/ADAS * 6 products. Furthermore, the two companies will continue to support the ever-evolving needs of mobility through the development of semiconductor technology.
< Comments from both companies >
Hiroshi Kondo, Head of Mobility Electronics Business Group, DENSO:
Through the use of system development capabilities and the combination of a broad range of technologies, DENSO will advance systems that connect vehicles with society at large. By doing so, we will tackle complex social issues moving forward, including achieving a carbon-neutral society, fostering a circular economy, and eliminating traffic accident fatalities.
One of the businesses we are strengthening is semiconductors, which is the foundation of the computing infrastructure for the advancement of vehicle intelligence. We have high expectations for the development of in-vehicle semiconductor IP that can flexibly respond to the AI trend with low power by utilizing quadric Chimera GPNPU.
Quadric.inc Veerbhan Kheterpal, CEO:
As a marquee Tier 1 automotive supplier throughout the world, DENSO subjected our technology to a comprehensive and intense benchmarking exercise.
Being selected by DENSO as their ADAS/AD processor IP partner is a great validation of the growing maturity and breadth of Quadric's GPNPU processors.
*1Neural Processing Unit (NPU): A processor specialized for AI computation
*2Chimera GPNPU (Chimera General Purpose Neural Processing Unit): The name of Quadric's processor architecture
*3SoC (System on a Chip): An integrated circuit that integrates multiple functions and functions as a system
*4RISC-V: An open standard Instruction Set Architecture (ISA).
*5ASIL (Automotive Safety Integrity Level)
*6AD/ADAS (Autonomous Driving/Advanced Driver Assistance System)
About DENSO CORPORATION
Globally headquartered in Kariya, Japan, DENSO is a $47.2 billion leading mobility supplier that develops advanced technology and components for nearly every vehicle make and model on the road today. With manufacturing at its core, DENSO invests in around 180 facilities worldwide to provide opportunities for rewarding careers and to produce cutting-edge electrification, powertrain, thermal and mobility electronics products, among others, that change how the world moves. In developing such solutions, the company's 162,000 global employees are paving the way to a mobility future that improves lives, eliminates traffic accidents, and preserves the environment. DENSO spent around 7.7 percent of its global consolidated sales on research and development in the fiscal year ending March 31, 2024. For more information about DENSO's operations worldwide, visitāhttps://www.denso.com/global.
www.prnewswire.com
DENSO and U.S. Startup Quadric Sign Development License Agreement for AI Semiconductor (NPU)
News provided by
DENSOOct 29, 2024, 12:41 ET
ā Collaboration Accelerates Joint Development of Semiconductor IP Suitable for Automotive Applications ā
KARIYA, Japan, Oct. 29, 2024 /PRNewswire/ -- DENSO CORPORATION and Quadric.inc have signed a development license agreement for a Neural Processing Unit (NPU) * 1, which is a semiconductor specialized for the arithmetic processing of AI. Through the agreement, DENSO will acquire the IP core license for Quadric's Chimera GPNPU * 2, and the two companies will co-develop intellectual property (NPU) for an in-vehicle semiconductor.
With the advancement of intelligent systems, such as automated driving and connected technologies, vehicles need to process in real time a vast amount of information obtained from on-board sensors, as well as inter-vehicle and cloud communications. For this reason, the performance of the in-vehicle SoC* 3, which can process large amounts of information at high speed and with efficiency, has become increasingly important.
DENSO and Quadric have been studying semiconductor IP (NPU) development using Chimera GPNPU (General Purpose Neural Processing Unit) for in-vehicle SoCs and have now decided to jointly develop an in-vehicle semiconductor IP (NPU) combining DENSO 's RISC-V* 4 based processor and Quadric's Chimera GPNPU.
DENSO 's RISC-V-based processor IP is compliant with ISO26262 ASIL * 5 D and is ideal for automotive applications, where safety is critical. As the intelligence of vehicles advances, AI will become a pivotal factor in determining the value of vehicles. Quadric's Chimera GPNPU has a unique architecture that can handle matrix and vector operations and scalar (control) code in one execution pipeline. This flexibility in handling diverse computational tasks makes it applicable to various AI functions. In addition, by using Chimera GPNPU, automotive system developers can add proprietary AI features to their products throughout the future.
By pursuing joint development of a new semiconductor IP (NPU), DENSO and Quadric aim to contribute to the realization of in-vehicle SoCs that can flexibly respond to changes in AI trends over a long period of time, from product development to post-release, and to promote the development of highly secure AD/ADAS * 6 products. Furthermore, the two companies will continue to support the ever-evolving needs of mobility through the development of semiconductor technology.
< Comments from both companies >
Hiroshi Kondo, Head of Mobility Electronics Business Group, DENSO:
Through the use of system development capabilities and the combination of a broad range of technologies, DENSO will advance systems that connect vehicles with society at large. By doing so, we will tackle complex social issues moving forward, including achieving a carbon-neutral society, fostering a circular economy, and eliminating traffic accident fatalities.
One of the businesses we are strengthening is semiconductors, which is the foundation of the computing infrastructure for the advancement of vehicle intelligence. We have high expectations for the development of in-vehicle semiconductor IP that can flexibly respond to the AI trend with low power by utilizing quadric Chimera GPNPU.
Quadric.inc Veerbhan Kheterpal, CEO:
As a marquee Tier 1 automotive supplier throughout the world, DENSO subjected our technology to a comprehensive and intense benchmarking exercise.
Being selected by DENSO as their ADAS/AD processor IP partner is a great validation of the growing maturity and breadth of Quadric's GPNPU processors.
*1Neural Processing Unit (NPU): A processor specialized for AI computation
*2Chimera GPNPU (Chimera General Purpose Neural Processing Unit): The name of Quadric's processor architecture
*3SoC (System on a Chip): An integrated circuit that integrates multiple functions and functions as a system
*4RISC-V: An open standard Instruction Set Architecture (ISA).
*5ASIL (Automotive Safety Integrity Level)
*6AD/ADAS (Autonomous Driving/Advanced Driver Assistance System)
About DENSO CORPORATION
Globally headquartered in Kariya, Japan, DENSO is a $47.2 billion leading mobility supplier that develops advanced technology and components for nearly every vehicle make and model on the road today. With manufacturing at its core, DENSO invests in around 180 facilities worldwide to provide opportunities for rewarding careers and to produce cutting-edge electrification, powertrain, thermal and mobility electronics products, among others, that change how the world moves. In developing such solutions, the company's 162,000 global employees are paving the way to a mobility future that improves lives, eliminates traffic accidents, and preserves the environment. DENSO spent around 7.7 percent of its global consolidated sales on research and development in the fiscal year ending March 31, 2024. For more information about DENSO's operations worldwide, visitāhttps://www.denso.com/global.
DENSO and U.S. Startup Quadric Sign Development License Agreement for AI Semiconductor (NPU)
/PRNewswire/ -- DENSO CORPORATION and Quadric.inc have signed a development license agreement for a Neural Processing Unit (NPU) * 1, which is a semiconductor...
Bravo
If ARM was an arm, BRN would be its bicepsšŖ!
Here is the information on MegaChips website on the different offerings between BrainChip and Quadric. Doesn't say anything about compatibility.
Image Processing | MegaChips Corporation
We respond to various issues of customers including cost and development period with our integrated support leveraging foundries around the world.
www.megachips.co.jp
Last edited:
Bravo
If ARM was an arm, BRN would be its bicepsšŖ!
Quiltman
Regular
Bravo, thankyou for the extensive material you publish on the forum.
It is for each of us to read, digest and make our personal judgement as to the impact on our investment ... but the fact that you & other key contributors publish regularly that makes the forum worthwhile.
Big picture for me is very positive.
Four to five years ago the debate was if neuromorphic architecture had a commercial future, or would be resigned to university laboratories.
The famous, "What is a neuromorphic chip anyway - wanca" , comment.
As to BrainChips share of that commercial pie .... it was hardly relevant if there was no pie to divvy up.
That has been put to rest.
There is now world wide consensus that neuromorphic architecture will be an essential part of the IT landscape - the discussion if about how many 10's, 100's or even 1000's of billions of $'s - and how quickly the market will adopt & realise this value.
Of course when the prize becomes so large it attracts a multitude of companies/investments/opinions. That is to be welcomed ! I would rather that than have Brainchip trying to scratch 20% share of a $500M market.
Will Brainchip become part of the "commercial" neuromorphic revolution. It would seem to me that the body of evidence suggests the answer is a resounding "yes", and that is why I remain invested, and continue to buy. Will it be this 4C - who the heck knows - but I remain extremely positive.
Thanks again.
Quiltman.
It is for each of us to read, digest and make our personal judgement as to the impact on our investment ... but the fact that you & other key contributors publish regularly that makes the forum worthwhile.
Big picture for me is very positive.
Four to five years ago the debate was if neuromorphic architecture had a commercial future, or would be resigned to university laboratories.
The famous, "What is a neuromorphic chip anyway - wanca" , comment.
As to BrainChips share of that commercial pie .... it was hardly relevant if there was no pie to divvy up.
That has been put to rest.
There is now world wide consensus that neuromorphic architecture will be an essential part of the IT landscape - the discussion if about how many 10's, 100's or even 1000's of billions of $'s - and how quickly the market will adopt & realise this value.
Of course when the prize becomes so large it attracts a multitude of companies/investments/opinions. That is to be welcomed ! I would rather that than have Brainchip trying to scratch 20% share of a $500M market.
Will Brainchip become part of the "commercial" neuromorphic revolution. It would seem to me that the body of evidence suggests the answer is a resounding "yes", and that is why I remain invested, and continue to buy. Will it be this 4C - who the heck knows - but I remain extremely positive.
Thanks again.
Quiltman.
Great post Quiltman, couldn't agree more, keep up your fantastic work Bravo, love your posts. Also don't care if the 4c is not all that great this time round, it's coming and I'm still holding and accumulating.
HopalongPetrovski
I'm Spartacus!
Whilst the band at BrainChip plays on, our Brava, Bravis! Bravo keeps Bravo-ing along.
God bless her cotton sox, painted tootsie's, spa parties, garage sales and elf boots.
She's the goat!
God bless her cotton sox, painted tootsie's, spa parties, garage sales and elf boots.
She's the goat!
Mazewolf
Regular
"Brainchip announces the new ultra low power Akida Kados neuromorphic chip and cortical hub architecture, optimised to massively reduce power consumption in data centres and servers. Applying edge power management strategies at scale, the Akida Kados aims to revolutionise power use globally"
ok now for the 4C....
Similar threads
