BRN Discussion Ongoing
- Thread starter TechGirl
- Start date
Tothemoon24
Top 20
Stephan Sokolov, Software Engineer at DeGirum, demonstrates the company’s latest edge AI and vision technologies and products in BrainChip’s booth at the 2025 Embedded Vision Summit. Specifically, Sokolov demonstrates the power of real-time AI inference at the edge, running DeGirum’s PySDK application directly on BrainChip hardware.
This demo showcases low-latency, high-efficiency performance as a script performs live inference on a video stream. Sokolov also highlights the DeGirum AI Hub—a cloud-based platform that allows developers to evaluate models and test deployments.
"We have integrated Akida into our PySDK and AI Hub." ... " running DeGirum’s PySDK application directly on BrainChip hardware"
So, on the PCIe board, the grey square below the asterisk is Akida
This is a nude Akida, without its flashy outer case. Bottom right is the characteristic chamfered corner. So is this from a new production run, or from the first engineering samples, or just an unboxed "commercial run" chip - or maybe 1500 (remembering that PCIe is not the mother board, so the CPU is on a different board)?
Here's a link to DeGirum's AI Hub:
https://www.prnewswire.com/news-rel...accelerate-edge-ai-development-302346154.html
SANTA CLARA, Calif., Jan. 13, 2025 /PRNewswire/ -- DeGirum®, a leader in edge AI software, announced the release of its AI Hub, an accessible evaluation platform designed to simplify AI prototyping and development at the edge. DeGirum's AI Hub enables developers to experiment with AI models for their applications. The platform supports a range of accelerators and application processors, including Hailo®, Intel®, MemryX®, BrainChip®, Rockchip®, and Google®.
"We are excited to launch our AI Hub which allows our customers to try many different industry leading AI processors and accelerators. 1000+ models are immediately available to try, with a wide range of algorithms and hardware solutions," said Bill Eichen, VP of Business Development at DeGirum.
DeGirum AI Hub
DeGirum's AI Hub enables rapid AI prototyping by providing access to hardware through an evaluation platform. The platform offers extensive model support, giving developers access to a diverse library of machine learning models to kickstart development. It features a model porting toolchain that converts training checkpoints into deployable models optimized for supported hardware targets. The AI Hub also hosts a model zoo, enabling edge devices to load and update models on demand, eliminating the need to bundle models within applications.
https://degirum.us.auth0.com/login?state=hKFo2SAycm5ubGxCU25EdnRwWmVhc0ZmTUlQdlRqUzVNVDB3UKFupWxvZ2luo3RpZNkgckw0RllpSkExSXJVV3p5YllJdWNUNkttMmhlcWdFUFCjY2lk2SBBQWxoeUltNFpxWWVBbEhQdjlqSnp0b09yRkhWQTNLYg&client=AAlhyIm4ZqYeAlHPv9jJztoOrFHVA3Kb&protocol=oauth2&scope=openid profile email offline_access&redirect_uri=https://hub.degirum.com&audience=https://degirum.us.auth0.com/api/v2/&response_type=code&response_mode=query&nonce=WTNhdUFRekxuTWRkRmdCMThaQ25OazROa0g1M0J2X2ZOTUJYaTQwOWJzWQ==&code_challenge=ZR0xgT2SjDRFBkZyZajqQt95XXbu5let9B_I7JUZ7L8&code_challenge_method=S256&auth0Client=eyJuYW1lIjoiYXV0aDAtdnVlIiwidmVyc2lvbiI6IjIuMy4xIn0=
Supported Hardware
Frangipani
Top 20
Laurent Hili on the ESA-supported BrainChip -Frontgrade Gaisler partnership:
www.linkedin.com
The partnership BrainChip - Frontgrade Gaisler is a significant step toward deploying AI in space. | Laurent Hili
The partnership BrainChip - Frontgrade Gaisler is a significant step toward deploying AI in space. With growing demand for higher processing capabilities, there is an emerging need for autonomous systems that can process data locally and make timely decisions. GR801 is a promising development...
Frangipani
Top 20
Akida gets a favourable mention alongside Loihi in a recent paper (“From Tiny Machine Learning to Tiny Deep Learning: A Survey”) by Texas State University researchers:
“[Neuromorphic] architectures promise efficient always-on inference on MCU-scale devices ideal for continuous monitoring applications-with hardware platforms like BrainChip’s Akida and Intel’s Loihi leading the way.”
(…)
“[Neuromorphic] architectures promise efficient always-on inference on MCU-scale devices ideal for continuous monitoring applications-with hardware platforms like BrainChip’s Akida and Intel’s Loihi leading the way.”
(…)
The key to using AI Chat boxes eg Gemini, co pilot etc is understanding that they have great long term memory but very poor short term memory.
It is hugely expensive to build short term memory in LLMs. So it's limited.
If you extend your conversation it will forget important details from early on in the conversation. Hence errors.
Questions need to be precise, explicit and provide context.
If the conversation goes on details from earlier may need to be resupplied in case it's forgotten.
If you want more detail on this ask your chat box and it will fill you in on the best way to use it.
Ask your chatbox. Are you always reliable with the infornation you supply.
It will tell you as it is.
It will even tell you when to be cautious etc.
Ask it. Are you smart.
Bravo
If ARM was an arm, BRN would be its biceps💪!
Here's a new Analog Devices job ad (posted 7 days ago) for an AI Compute Architect being hired to lead architectural integration of µAI (micro AI) compute blocks, analog AI, in-memory computing and spiking neural networks / neuromorphic computing.
A pretty clear signal IMO that neuromorphic approaches are on ADI’s SoC roadmap.
As you can see, they’re looking for someone to help with early RTL and analog prototyping, interface definition and IP integration, which shows they’re not just researching but planning to potentially integrate third-party IPs (like BrainChip’s Akida or some other similar neuromorphic technology).
The job responsibilities include:
“Support early RTL and analog prototyping of both internal and third-party IP blocks.”
That's a pretty strong indicator they’re at least open to collaborating with companies like BrainChip.
Obviously, it's by no means confirmation of any sort of partnership, but it’s certainly a door that seems to be open.
No need to wait on the welcome mat, Sean. Cruise on through like you mean business!
A pretty clear signal IMO that neuromorphic approaches are on ADI’s SoC roadmap.
As you can see, they’re looking for someone to help with early RTL and analog prototyping, interface definition and IP integration, which shows they’re not just researching but planning to potentially integrate third-party IPs (like BrainChip’s Akida or some other similar neuromorphic technology).
The job responsibilities include:
“Support early RTL and analog prototyping of both internal and third-party IP blocks.”
That's a pretty strong indicator they’re at least open to collaborating with companies like BrainChip.
Obviously, it's by no means confirmation of any sort of partnership, but it’s certainly a door that seems to be open.
No need to wait on the welcome mat, Sean. Cruise on through like you mean business!
AI Compute Architect
About Analog Devices Analog Devices, Inc. (NASDAQ: ADI ) is a global semiconductor leader that bridges the physical and digital worlds to enable breakthroughs at the Intelligent Edge. ADI combines analog, digital, and software technologies into solutions that help drive advancements in digitized...
analogdevices.wd1.myworkdayjobs.com
Last edited:
Morning Chippers,
Nothing on us , .....yet .
Meta buying a little stake .... Smart Glasses.
Shares of Ray-Ban maker EssilorLuxottica jump on reported Meta stake https://share.google/3o9s29xk9X9uwZDJV
Regards,
Esq.
Nothing on us , .....yet .
Meta buying a little stake .... Smart Glasses.
Shares of Ray-Ban maker EssilorLuxottica jump on reported Meta stake https://share.google/3o9s29xk9X9uwZDJV
Regards,
Esq.
Bravo
If ARM was an arm, BRN would be its biceps💪!
NVIDIA is on 
www.abc.net.au
Nvidia becomes first listed company with $US4 trillion valuation
Amid an investor frenzy around artificial intelligence, the valuation for Silicon Valley chipmaker Nvidia reaches $US4 trillion ($6.12 trillion).
And we have what they need to get more efficient! And they know it!NVIDIA is on
Nvidia becomes first listed company with $US4 trillion valuation
Amid an investor frenzy around artificial intelligence, the valuation for Silicon Valley chipmaker Nvidia reaches $US4 trillion ($6.12 trillion).
It’s about time !
Baron Von Ricta
Regular
Manny, thanks for your insight. Sounds a bit like me - limited short term memory!
Wow so this is how a working modern crystal ball looks like… impressive! I’m still old school man… still manual…
Share price being manipulated because there is no revenue nor material news on ASX. It occurs everywhere, and unfortunately, we have a strong track record of no material news for agesShareholders,
Do you believe that our technology has advanced over the last 3 years or not ?
I think yes, you think what ?
Shareholders,
Do you think our website has improved, as in, overall information sharing ?
I think yes, you think what ?
Shareholders,
Do you believe that Antonio lied during the AGM ?
I think yes, you think what ?
Shareholders,
Do you believe that Antonio's comments at the AGM were addressed by fellow Board members after the conclusion of the AGM ?
I think yes, you think what ?
Shareholders,
Do you believe that Sean, who is 3 years 9 months into his tenure will deliver before his 5 year (agreement) is due to expire ?
I think yes, you think what ?
Shareholders,
Do you believe that the Board will suggest that we re-domicle to the US in early 2026 ?
I think not, you think what ?
Shareholders,
Are you happy with how our share price is being manipulated by an automated robotic algorithm, to the point of artificial control ?
I think not, you think what ?
Shareholders,
Do you believe that Tony is the right man to be leading our technology team ?
I think yes, you think what ?
Just some honest, genuine questions that may confirm why you are still invested in our company.
The Akida family, we are growing, please don't loose sight of that FACT !
Tech x
Pom down under
Top 20
Strong don’t you meanwe have a strong track record of no material news for ages
Tothemoon24
Top 20
GlobalFoundries buys MIPS for edge AI network dominion
The chip maker banks on the promise of RISC-VJuly 09, 2025 By Giacomo "Jack" Lee Comment
FacebookTwitterLinkedInRedditEmailShare
GlobalFoundries announced a definitive agreement to acquire chip design company MIPS, in a move focused around AI-enabled network infrastructure.
In its announcement from Tuesday, the New York-headquartered foundry firm shared that MIPS will continue to operate as a standalone business within the new setup. That business revolves around developing processor architectures, now complementing GlobalFoundries’ (GF) semiconductor manufacturing capabilities.
What makes MIPS particularly attractive to GF is its relevance in AI chip design. Recent years have seen the Sunnyvale, California business focus on RISC-V, poaching execs from the likes of AMD and RISC-V rival SiFive to bolster its mission.
Pronounced risk-five, RISC-V is an open standard instruction set (ISA) architecture based on established reduced instruction set computing (RISC) principles. Based on this, the firm’s Atlas portfolio offers compute cores designed for real-time, low-latency processing and, perhaps most pertinently, edge AI processing cores.
With the acquisition, GF is in a strong position to offer custom chip solutions for smart routers, switches, and edge-network devices that incorporate AI and machine learning functions closer to the network’s edge.
With in-house processor IP and fabrication, its custom chips are likely to attract the likes of network equipment vendors such as Cisco, and OEMs like Nokia for SoCs in 5G small cells and base stations.
AI on the edge
The low-latency and high bandwidth efficiency of AI directly embedded in edge devices is driving more interest in the market.As reported by SDxCentral, most initial investment in edge AI is targeted at equipment that enterprises use to support on-prem use cases.
Companies like Verizon have already signed deals with the likes of Nvidia to power enterprise AI services. While ROI remains low and IoT persists as the main edge use case driver, IDC predicts the edge compute market is set for an AI-fueled boost pushing overall segment spend to $380 billion by 2028.
With MIPS on board, GF finds itself in a good position for all this loot. That said, how big of a deal the acquisition is has not been disclosed. The last public record of a MIPS sale was for $60 million, when Tallwood Venture Capital bought the company from Imagination Technologies in 2017.
Since then, the company has changed hands a few times, being bought out by Wave Computing in 2018 for an undisclosed amount.
According to Reuters, Shanghai’s CIP United acquired full licensing rights to MIPS architecture for mainland China and its territories in 2019 for $60 million. Those rights for China will remain in CIP's hands in spite of GF’s global operations of MIPS.
GF’s acquisition of MIPS is expected to close in the second half of 2025.
GlobalFoundries buys MIPS for edge AI network dominion
The chip maker banks on the promise of RISC-V
July 09, 2025 By Giacomo "Jack" Lee Comment
FacebookTwitterLinkedInRedditEmailShare
Sundry Photography/Getty Images
GlobalFoundries announced a definitive agreement to acquire chip design company MIPS, in a move focused around AI-enabled network infrastructure.
In its announcement from Tuesday, the New York-headquartered foundry firm shared that MIPS will continue to operate as a standalone business within the new setup. That business revolves around developing processor architectures, now complementing GlobalFoundries’ (GF) semiconductor manufacturing capabilities.
What makes MIPS particularly attractive to GF is its relevance in AI chip design. Recent years have seen the Sunnyvale, California business focus on RISC-V, poaching execs from the likes of AMD and RISC-V rival SiFive to bolster its mission.
Pronounced risk-five, RISC-V is an open standard instruction set (ISA) architecture based on established reduced instruction set computing (RISC) principles. Based on this, the firm’s Atlas portfolio offers compute cores designed for real-time, low-latency processing and, perhaps most pertinently, edge AI processing cores.
With the acquisition, GF is in a strong position to offer custom chip solutions for smart routers, switches, and edge-network devices that incorporate AI and machine learning functions closer to the network’s edge.
With in-house processor IP and fabrication, its custom chips are likely to attract the likes of network equipment vendors such as Cisco, and OEMs like Nokia for SoCs in 5G small cells and base stations.
AI on the edge
The low-latency and high bandwidth efficiency of AI directly embedded in edge devices is driving more interest in the market.
As reported by SDxCentral, most initial investment in edge AI is targeted at equipment that enterprises use to support on-prem use cases.
Companies like Verizon have already signed deals with the likes of Nvidia to power enterprise AI services. While ROI remains low and IoT persists as the main edge use case driver, IDC predicts the edge compute market is set for an AI-fueled boost pushing overall segment spend to $380 billion by 2028.
With MIPS on board, GF finds itself in a good position for all this loot. That said, how big of a deal the acquisition is has not been disclosed. The last public record of a MIPS sale was for $60 million, when Tallwood Venture Capital bought the company from Imagination Technologies in 2017.
Since then, the company has changed hands a few times, being bought out by Wave Computing in 2018 for an undisclosed amount.
According to Reuters, Shanghai’s CIP United acquired full licensing rights to MIPS architecture for mainland China and its territories in 2019 for $60 million. Those rights for China will remain in CIP's hands in spite of GF’s global operations of MIPS.
GF’s acquisition of MIPS is expected to close in the second half of 2025.
Yeah but you know….
“We are pushing the boundaries to enable AI on the edge.”
Global Foundries are the makers of the Akida 1500 22nm FD-SoI chips.
https://brainchip.com/brainchip-rec...icon-from-technology-partner-globalfoundries/
It looks like the wisdom of our early engagement with RISC-V company SiFive is beginning to shine through.
Chelpis, Andes, Frontgrade all use RISC-V.
MIPS have a suite of patents relating to a switching fabric array for interconnecting NNs. It's possible that this is used in Akida to select the NPU arrangement for the layers of a NN.
US2020034262A1 PROCESSOR ARRAY REDUNDANCY 20180727
processor synchronization within a reconfigurable computing environment for processor array redundancy. Processing elements are configured within a reconfigurable fabric to implement two or more redundant processors, where the two or more redundant processors are enabled for coincident operation. An agent is loaded on each of the two or more redundant processors, where the agent performs a function requiring data validation. The agent is fired on each of the two or more redundant processors to commence coincident operation. The coincident operation can include a lockstep operation. An output data result from each of the two or more redundant processors is compared to enable a data validation result. The data validation result is propagated. The propagating the data validation result can be based on comparing valid output data or can be based on comparing invalid output data.
So maybe BRN was the marriage broker?
https://brainchip.com/brainchip-rec...icon-from-technology-partner-globalfoundries/
It looks like the wisdom of our early engagement with RISC-V company SiFive is beginning to shine through.
Chelpis, Andes, Frontgrade all use RISC-V.
MIPS have a suite of patents relating to a switching fabric array for interconnecting NNs. It's possible that this is used in Akida to select the NPU arrangement for the layers of a NN.
US2020034262A1 PROCESSOR ARRAY REDUNDANCY 20180727
processor synchronization within a reconfigurable computing environment for processor array redundancy. Processing elements are configured within a reconfigurable fabric to implement two or more redundant processors, where the two or more redundant processors are enabled for coincident operation. An agent is loaded on each of the two or more redundant processors, where the agent performs a function requiring data validation. The agent is fired on each of the two or more redundant processors to commence coincident operation. The coincident operation can include a lockstep operation. An output data result from each of the two or more redundant processors is compared to enable a data validation result. The data validation result is propagated. The propagating the data validation result can be based on comparing valid output data or can be based on comparing invalid output data.
So maybe BRN was the marriage broker?
Pom down under
Top 20
Bravo
If ARM was an arm, BRN would be its biceps💪!
Sounds very promising @Diogense!
It says in the article that @Toothemoon posted:
"With the acquisition, GF is in a strong position to offer custom chip solutions for smart routers, switches, and edge-network devices that incorporate AI and machine learning functions closer to the network’s edge."
Talk about being right down our alley! Surely BrainChip would be a great fit as a complementary IP block into MIPS-designed RISC-V SoCs for reducing power and latency in edge ML workloads, enabling local learning, offloading inference, etc.
Prior fab partner for BrainChip - tick
Synergy with RISC-V (i.e. Si-Five, etc. ) - tick
Edge AI focus - tick
GF is pushing into custom SoC development with vertical stacks - tick
If BrainChip wanted to tape out Akida 2 or Akida 3 in silicon, GF would be a natural foundry candidate.
Last edited:
Similar threads
