CHIPS
Regular
Nothing new, it is the ONSOR glasses.
BRN Discussion Ongoing
- Thread starter TechGirl
- Start date
Nothing new, it is the ONSOR glasses.
Fullmoonfever
Top 20
Just managed to skim some of the flurry of posts today and nice to see an Ann at least.
The 3 that jump out at me from memory, who have been already linked to the 1500 as partners:
Onsor.
Andes.
HaiLa.
We know if all goes well for Onsor that they want to be hitting the mkt with a product next year.
Do the 3 above go with IP or is it easier / better for them to have physical 1500s?
The 3 that jump out at me from memory, who have been already linked to the 1500 as partners:
Onsor.
Andes.
HaiLa.
We know if all goes well for Onsor that they want to be hitting the mkt with a product next year.
Do the 3 above go with IP or is it easier / better for them to have physical 1500s?
Space Cadet
Regular
Bascom Hunter doubling production facilities in Baton Rogue. 3 days old. Can't find any details. Behind pay wall.
https://www.opportunitylouisiana.go...uge-footprint-with-a-new-fabrication-facility
Wonder why?
SC
Last edited:
Aha! Thereby hangs a tail - "there by what hangs a tail?" you may well ask.
The cost of implementing an IP licence would exclude most small/medium enterprises.
Taproot
Regular
+ VVDN
Future VVDN and BrainChip boxes could use BrainChip’s second-gen AKD1500, which is more compact and efficient, Telson said.
+ Bascom Hunter
In late December, BrainChip signed a US$100,000 contract with U.S. defence contractor Bascom Hunter, for the sale and support of AKD1500 chips for full scale evaluation of commercial products, which will provide significant improvements over traditional CPU, GPU, and TPU processors
More than 2 years since AKD1500 Version 1
Another 12 months for Version 2.
Certainly does test ones patience.
BrainChip Receives First Shipment of AKD1500 Chips in Silicon from Technology Partner GlobalFoundries
BrainChip's latest milestone: AKD1500 chip arrival from GlobalFoundries. Explore the future of semiconductors here.
brainchip.com
Bravo
If ARM was an arm, BRN would be its biceps💪!
Hi JD,
For what it's worth, I asked ChatGPT for a neutral take on this.
What I thought was interesting, was that it weaved in Anduril's headset into the response. I didn't prompt it to, but earlier in the day, in a separate chat, I had queried ChatGPT on boards or integration possibilities and whether an Akida 1500 + Arm's Cortex M85 module (as an example) could power something like Anduril’s helmet.
ChatGPT's Response
What “supports 128 MACs per node” most likely means
- Each Akida node has a local 4×4 MAC micro-engine (128 MACs) it can invoke for certain ops.
- Those MACs co-exist with the spiking compute (CNP/FNP or equivalent). They’re there so models can include non-spiking layers/ops without leaving the chip.
Why add MACs if you have an SNN?
Because many modern edge models are hybrid. MACs cover things SNNs don’t excel at (or that are simpler to drop in as-is):- Conv/pointwise (1×1) / depthwise conv blocks
- Linear/projection layers (e.g., classifier heads, QKV in tiny attention)
- Normalisation / pooling / residual glue
- Pre/post-processing (e.g., short FIRs, feature projections)
- Direct reuse of existing INT4/INT8 CNN chunks without full SNN conversion
What a hybrid pipeline might look like in a helmet
Sensors- Front/rear frame cameras → light MAC preproc (resize/1×1 conv).
- Event camera/radar/audio → feed SNN/TENNs directly.
- Perception
- SNN handles temporal/gesture/track (event streams).
- MAC handles small CNN blocks or projection heads.
- Fusion & decisions on the host MCU (e.g., Cortex-M85).
Bottom line
- The “128 MACs per node” doesn’t signal the end of Akida’s neuromorphic core. It signals a pragmatic hybrid: keep the spiking engine for temporal, sparse, event-driven strengths, and use local MACs to natively run conventional layers (conv/linear/norm) and reuse existing model chunks.
- That combo generally improves accuracy, eases porting, and lowers total latency/power versus forcing everything into SNN or everything into classic CNN on a separate chip.
Tothemoon24
Top 20
Thanks Sean
Fullmoonfever
Top 20
Who knows...maybe the 1500 is something Bascom would look at but at the mo they appear to have really expanded and outlined their Akida Snap Card offering details since I looked some time ago.+ VVDN
Future VVDN and BrainChip boxes could use BrainChip’s second-gen AKD1500, which is more compact and efficient, Telson said.
+ Bascom Hunter
In late December, BrainChip signed a US$100,000 contract with U.S. defence contractor Bascom Hunter, for the sale and support of AKD1500 chips for full scale evaluation of commercial products, which will provide significant improvements over traditional CPU, GPU, and TPU processors
More than 2 years since AKD1500 Version 1
Another 12 months for Version 2.
Certainly does test ones patience.
BrainChip Receives First Shipment of AKD1500 Chips in Silicon from Technology Partner GlobalFoundries
BrainChip's latest milestone: AKD1500 chip arrival from GlobalFoundries. Explore the future of semiconductors here.
Comes under their mixed signal product line.
The Mixed Signal Product line is currently focused on VPX based solutions – especially those that require both digital and analog or photonic interfaces. Our VPX Cards include 3rd party VPX cards with Bascom Hunter firmware, custom VPX cards and support electronics such as Mezzanine cards to provide additional functionality to VPX users. Our VPX Chassis include custom built chassis and custom backplanes.
3U VPX SNAP Card
bascomhunter.com
Bascom Hunter’s SNAP Card (Spiking Neuromorphic Advanced Processor) is a high-performance 3U OpenVPX AI/ML processor built for rugged, mission-critical environments. The card is SOSA-aligned, HOST-compatible, and conduction cooled. It combines a Xilinx UltraScale+ RFSoC FPGA with five BrainChip AKD1000 spiking neuromorphic processors to achieve the best of both signal processing and neuromorphic computing – featuring a total of 6 million neurons and 50 billion synapses across the card. Unlike traditional machine learning accelerators such as GPUs or TPUs, neuromorphic processors are designed to mimic the biological efficiency of the human brain, allowing the SNAP Card to run multiple ML models in parallel at exceptionally low power — just 1W per model — without sacrificing speed or accuracy while the FPGA enhances input/output operations and performs any desired signal processing tasks. This combination in a rugged, interoperable, and military-hardened package makes Bascom Hunter’s SNAP Card the ideal solution for the concurrent and parallel processing of real-time, multi-modal, and multi sensor data on autonomous, unattended, denied, or otherwise battery constrained military systems.
Download Brochure
APPLICATION
A single SNAP Card, deployed on an airborne ISR platform, occupying only one 3U VPX slot, could deploy up to five complex machine learning algorithms across each of its onboard processors to perform the following missions simultaneously.- Automated Target Recognition (ATR) on full motion video feeds and imagery (to include 4k and higher)
- Real-time detection and identification of threat radars and their acquisition/operating modes
- Detection of FISINT (Foreign Instrumentation Signature Intelligence) and/or hacking across the airframe’s 1553 communications bus
- Perform communications analysis to include speech-to-text and foreign language translation of intercepted communications
- Fuse multiple infrared cameras (e.g., SWIR, MWIR, LWIR) to provide a combined infrared operating picture on the ground Multi-user, simultaneous modulation/demodulation
MEMORY
|
|
SPIKING NEURAL NETWORK PROCESSORS
|
|
INTEGRATION SPECIFICATIONS
|
|
VPX FORM FACTOR
|
|
Loved the earlier shorts will burn gifs, Idiots, today’s announcement was about management saying don’t worry shorters we got you covered, you don’t have to close your 60 plus million shorts yet, Capital raise coming we will look after you sophisticated investors again and let you close by gifting you cheap shares. What a load of wankers our managers and directors are, Played all retail investors for fools, remember when the AGM comes what they have done by screwing us all yet again.
FJ-215
Regular
Hi @7für7 ,
"That’s why it makes perfect sense for us to offer both: licensing and hardware. It makes us more flexible and opens more doors ..both with existing and new partners.
What do you think?"
The original plan of offering both chips and IP made perfect sense and still does.
When covid hit, it opened the door for BRN in a way, we had their slots at TSMC and gave them up but not everybody did. That is what lead to some of the supply chain issues. Auto manufacturers had assembled cars they couldn't deliver because they cancelled their chip orders and were put at the back of the queue.
Unfortunately, some years ago, Antonio put us on a merry-go-round and decided we needed to "course correct", change horses mid-stream and put all our eggs in one basket. Now we are "pivoting" into a double backflip (maybe with a half twist and pike) and if we stick the landing, will be back to where we started from. Hopefully facing in the same direction and sprinting for the finishing line!
Edit...... or is that the starting line?
I had a conversation in the German forum where a well known user (also here some time ago) wrote this … and I would like to hear your opinion… translated in English
“Good morning, . A tape-out is not the same as mass production. The AKD1000 has had a tape-out for years, yet there’s still no mass production, because there are no orders.
The AKD1500 isn’t really an update — it’s a stripped-down version of the AKD1000. The difference is that the AKD1500 can’t run as a standalone chip; it only works in combination with a CPU. That basically cancels out the intended power savings.
So there’s definitely no reason for excitement here. In fact, this announcement mainly shows that BrainChip’s “IP-only” business model isn’t getting them anywhere. It doesn’t appear to have any real commercial background either.
If there were a confirmed order for the AKD1500, it would have to be announced anyway, since it would be price-sensitive information.”
“Good morning, . A tape-out is not the same as mass production. The AKD1000 has had a tape-out for years, yet there’s still no mass production, because there are no orders.
The AKD1500 isn’t really an update — it’s a stripped-down version of the AKD1000. The difference is that the AKD1500 can’t run as a standalone chip; it only works in combination with a CPU. That basically cancels out the intended power savings.
So there’s definitely no reason for excitement here. In fact, this announcement mainly shows that BrainChip’s “IP-only” business model isn’t getting them anywhere. It doesn’t appear to have any real commercial background either.
If there were a confirmed order for the AKD1500, it would have to be announced anyway, since it would be price-sensitive information.”
Of course I disagreed in some points…lots of points actually…and he responded this:
“Well, it has to get technical again — no other way around it. Overall, a move away from SNNs is becoming apparent. It’s actually a logical step, since there are no algorithms that can properly train SNNs, and at present no one can predict if or when they’ll become available.
This already began with the development of TENNs, which are a CNN-based upgrade. Because the AKD1500 requires a CPU, it can only be used as an accelerator within a CNN. But that then requires the troublesome CNN-to-SNN conversion, which leads to increased latency.
In the CNN field there are plenty of competitive solutions, and power consumption there continues to drop. With the introduction of 6G, the whole situation will be reshuffled anyway, as completely different amounts of data will be processed. Integrating neuromorphic chips into CNNs will always remain a compromise solution. Only a true SNN can really tap into the potential of neuromorphic chips.
But as mentioned, those pesky algorithms… Intel is working on suitable ones with its neuromorphic supercomputer Hala Point, but so far without success — and those resources are far beyond anything BrainChip has access to.
A commercial breakthrough certainly won’t come before 2030, but we’ll probably keep seeing the same old motivational slogans posted here over the next five years.”
Thanks for your opinions
“Well, it has to get technical again — no other way around it. Overall, a move away from SNNs is becoming apparent. It’s actually a logical step, since there are no algorithms that can properly train SNNs, and at present no one can predict if or when they’ll become available.
This already began with the development of TENNs, which are a CNN-based upgrade. Because the AKD1500 requires a CPU, it can only be used as an accelerator within a CNN. But that then requires the troublesome CNN-to-SNN conversion, which leads to increased latency.
In the CNN field there are plenty of competitive solutions, and power consumption there continues to drop. With the introduction of 6G, the whole situation will be reshuffled anyway, as completely different amounts of data will be processed. Integrating neuromorphic chips into CNNs will always remain a compromise solution. Only a true SNN can really tap into the potential of neuromorphic chips.
But as mentioned, those pesky algorithms… Intel is working on suitable ones with its neuromorphic supercomputer Hala Point, but so far without success — and those resources are far beyond anything BrainChip has access to.
A commercial breakthrough certainly won’t come before 2030, but we’ll probably keep seeing the same old motivational slogans posted here over the next five years.”
Thanks for your opinions
Last edited:
Anduril Lattice is conventional AI and still requires cloud support. They have Partnered with Palantir AI to alleviate some problems resulting from this.
Anduril and Palantir to Accelerate AI Capabilities for National Security
Palantir and Anduril, two leading companies at the intersection of commercial technology and national security, are launching a new consortium to ensure that the U.S. government leads the world in artificial intelligence.
So its very possible AKIDA is a superior solution or as a hybrid to cover all bases.
Are Onsor using the AKIDA 1000 or 1500 chip? Is there any confirmation??
Onsor - according to their website:
2022 - " Identification of potential applications for the technology and the discovery of "epilepsy prediction" as a primary application, marking the official start of the project."
onsor.om
BrainChip announced the AKD1500 reference design was taped out on January 29, 2023.
brainchip.com
During 2022 when the Epilepsy prediction was discovered i presume they would have had to use the AKIDA 1000 chips as the 1500 was not yet available.
"Phase 3 - Preliminary results from the research emerge."
" Phase 3.5 - Preparation of documentation for patent application."
They may have upgraded to the 1500 but this would mean they had to change horses during the process.
.
Onsor - according to their website:
2022 - " Identification of potential applications for the technology and the discovery of "epilepsy prediction" as a primary application, marking the official start of the project."
NEXA | Onsor
BrainChip Tapes Out AKD1500 Chip in GlobalFoundries 22nm FD SOI Process
BrainChip tapes out the AKD1500 chip using GlobalFoundries' 22nm FD-SOI process, marking a significant milestone in AI chip development.
brainchip.com
"Phase 3 - Preliminary results from the research emerge."
" Phase 3.5 - Preparation of documentation for patent application."
They may have upgraded to the 1500 but this would mean they had to change horses during the process.
.
FJ-215
Regular
What was your first clue?????
Oh, i assumed you must have missed it. Well done.
FJ-215
Regular
A poster earlier alerted us that Bascom Hunter were planning to double their Baton Rouge footprint with a new fabrication facility.
Bascom Hunter is doubling its Baton Rouge footprint with a new fabrication facility - LED | Louisiana Economic Development
Unfortunately behind a pay wall.
Interesting from BH website:
" Bascom Hunter produces three types of Neuromorphic Processors.
The first is the deployment of a neural network into an FPGA. Doing this allows us to take existing COTS hardware and help customers bring existing (or proposed) neural networks into this hardware, speeding up the computation and lowering the power requirements. The second is the deployment of a neural network into custom hardware – typically based on a 3rd party ASIC processor. We can take ASICs produced by the vendor of your choice and put it into a platform and mission specific form factor. For example, into VPX or small-sat form factors. The third type of Neuromorphic processor is intended for our future focused customers looking to engage on cutting edge photonic based neuromorphic processors."
" RAPID DEPLOYMENT OF ACTIONABLE INSIGHTS
FPGAs offer tangible improvements over traditional CPU and GPU processors for both sensor deployments and Machine Learning (ML) tasks. Bascom Hunter excels in partnering with military and intelligence partners to rapidly develop custom FPGA solutions that leverage SoC and neuromorphic processors that tackle some of the toughest challenges facing our nation today."
My bold above.
Neuromorphic Processors - Bascom Hunter
Bascom Hunter is doubling its Baton Rouge footprint with a new fabrication facility - LED | Louisiana Economic Development
Unfortunately behind a pay wall.
Interesting from BH website:
" Bascom Hunter produces three types of Neuromorphic Processors.
The first is the deployment of a neural network into an FPGA. Doing this allows us to take existing COTS hardware and help customers bring existing (or proposed) neural networks into this hardware, speeding up the computation and lowering the power requirements. The second is the deployment of a neural network into custom hardware – typically based on a 3rd party ASIC processor. We can take ASICs produced by the vendor of your choice and put it into a platform and mission specific form factor. For example, into VPX or small-sat form factors. The third type of Neuromorphic processor is intended for our future focused customers looking to engage on cutting edge photonic based neuromorphic processors."
" RAPID DEPLOYMENT OF ACTIONABLE INSIGHTS
FPGAs offer tangible improvements over traditional CPU and GPU processors for both sensor deployments and Machine Learning (ML) tasks. Bascom Hunter excels in partnering with military and intelligence partners to rapidly develop custom FPGA solutions that leverage SoC and neuromorphic processors that tackle some of the toughest challenges facing our nation today."
My bold above.
Neuromorphic Processors - Bascom Hunter
FJ-215
Regular
Hi manny,
All jokes aside, looking at the tech roadmap, we do have some big ticket items coming up that will require a good amount of fresh cash. The AKD1500 tape out is a plus one on top of that.
I don't see any of this as a negative!!!!
The market knows that we have the LDA agreement at our disposal to raise the capital required.
What I want to see is an impressive IP announcement to offset the next LDA call.
Not asking much!
Similar threads
