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:
Similar threads
