Bugger! I just remembered - Akida is one-shot learning!
BRN Discussion Ongoing
- Thread starter TechGirl
- Start date
Deadpool
Regular
You in that dress? The plot twist no one saw coming, see you at Mardi Gras darling, I,ll be the one wearing fishnets.
Baron Von Ricta
Regular
Baron Von Ricta
Regular
Sorry that should read "can't have". Guess I made a BRN announcement mistake. My apologies.Sorry Bravo, but no ones going to tell me I can't gave more beer.
Typos are a hangin' offence 'round
HopalongPetrovski
I'm Spartacus!
I thought everyone was already aware that I am a smoking hot sexbot with an insatiable lust for fat, bald and irascible old men who have an irrational love for BrainChip. Come, let me oil up that hair bereft noggin you reside in.
Tothemoon24
Top 20
Interesting read from Toshiba.
Didn’t we have a long bow link at some point
More to read here ;
athletechnews.com
Didn’t we have a long bow link at some point
More to read here ;
Toshiba Exec Breaks Down Emerging AI Trends in Retail
Toshiba Global Commerce Solutions is working to bring small and mid-sized retailers into the AI conversation.
athletechnews.com
TheDrooben
Pretty Pretty Pretty Pretty Good
Fortunately whisky is not kept in the fridge ehSorry Bravo, but no ones going to tell me I can't gave more beer.
Okeydokey01
Regular
To be fair they may very well be able to organise the piss up.They couldn’t even arrange a piss up in a brewery
The problem is the board would choose not to announce it so no one would turn up!...
Last edited:
Evening Tothemoon24 ,Interesting read from Toshiba.
Didn’t we have a long bow link at some point
View attachment 89394
View attachment 89393
More to read here ;
Toshiba Exec Breaks Down Emerging AI Trends in Retail
Toshiba Global Commerce Solutions is working to bring small and mid-sized retailers into the AI conversation.
Loose connection dating back to 15th March 2023 ,
Relating to a company announcement on a new partner , IPSOLON , thay also let slip Toshiba , Cisco & Linaro , then quickly deleted the last three.
Someone may have captured said announcement , before it was eddited.
Regards,
Esq.
Fullmoonfever
Top 20
This would be the screenshot mate care of @stuart888
BRN Discussion Ongoing
Renesas is the number three supplier of MCU’s to the automotive sector and well ahead of Bosch which everyone has likely heard of. However it seems clear that we need a motion at the AGM requiring Brainchip to cancel any IP licences not sold directly by Brainchip. Apparently there is something...
thestockexchange.com.au
No wonder ASX is dirty on tech (no relation):
https://www.abc.net.au/news/2025-08...boe-as-tpg-mishap-adds-10-year-woes/105623428
...
In 2015, the ASX began scouting for a replacement to the ageing technology it used to settle trades on the exchange.
Two years later, it created global headlines. In a market abuzz with talk of cryptocurrency and its open source ledger system, the ASX announced it would build the world's first industrial scale blockchain for financial services applications.
The timeline was always ambitious. It was supposed to be online by 2020.
But the project became ever-more complex as fights developed between various information providers about how they would interact with the new system.
Shares do not simply change hands between buyers and sellers — there are share registries, custodians and a host of other players, many of whom became concerned the new system would steal their business.
By the time the fifth delay to the rollout time was announced, it was obvious the project was on the rocks. At the end of 2022, it was canned, forcing the ASX to announce a $250 million write-off.
Brokers and investment houses had spent vast amounts too, replacing their systems to integrate with the blockchain dream that ultimately turned into a blocked drain.
Dominic Stevens, the ASX chief executive who commissioned the project, had left at the start of the year, leaving then chair Damian Roche to clean up the mess and to appoint Accenture to independently review what had gone wrong.
"On behalf of ASX, I apologise for the disruption experienced in relation to the CHESS replacement project over a number of years," he said at the time.
The original story was about lettingthe a Chicago mob set up a rival stock exchange to break ASX monopoly after a monumental cock up ...
" ASX faces losing virtual monopoly as TPG bungle adds to a decade of woes "
On Wednesday, the ASX confused a listed company with a similarly-named foreign owned private equity group that was engaged in a huge takeover.
The mistake resulted in TPG Telecom shares plummeting 5 per cent, wiping $400 million from its market value, even though it had nothing to do with the $645 million takeover of automotive software group Infomedia.
If the original mix-up was bad, the inability of the ASX to rectify the situation turned it into a debacle, as traders pounded TPG Telecom's stock for hours.
And it's unlikely to be the last the operator hears from TPG, with the telco understood to be considering its legal options.
So maybe we are moving to the Chicago exchange ... ?
https://www.abc.net.au/news/2025-08...boe-as-tpg-mishap-adds-10-year-woes/105623428
...
In 2015, the ASX began scouting for a replacement to the ageing technology it used to settle trades on the exchange.
Two years later, it created global headlines. In a market abuzz with talk of cryptocurrency and its open source ledger system, the ASX announced it would build the world's first industrial scale blockchain for financial services applications.
The timeline was always ambitious. It was supposed to be online by 2020.
But the project became ever-more complex as fights developed between various information providers about how they would interact with the new system.
Shares do not simply change hands between buyers and sellers — there are share registries, custodians and a host of other players, many of whom became concerned the new system would steal their business.
By the time the fifth delay to the rollout time was announced, it was obvious the project was on the rocks. At the end of 2022, it was canned, forcing the ASX to announce a $250 million write-off.
Brokers and investment houses had spent vast amounts too, replacing their systems to integrate with the blockchain dream that ultimately turned into a blocked drain.
Dominic Stevens, the ASX chief executive who commissioned the project, had left at the start of the year, leaving then chair Damian Roche to clean up the mess and to appoint Accenture to independently review what had gone wrong.
"On behalf of ASX, I apologise for the disruption experienced in relation to the CHESS replacement project over a number of years," he said at the time.
The original story was about letting
" ASX faces losing virtual monopoly as TPG bungle adds to a decade of woes "
On Wednesday, the ASX confused a listed company with a similarly-named foreign owned private equity group that was engaged in a huge takeover.
The mistake resulted in TPG Telecom shares plummeting 5 per cent, wiping $400 million from its market value, even though it had nothing to do with the $645 million takeover of automotive software group Infomedia.
If the original mix-up was bad, the inability of the ASX to rectify the situation turned it into a debacle, as traders pounded TPG Telecom's stock for hours.
And it's unlikely to be the last the operator hears from TPG, with the telco understood to be considering its legal options.
So maybe we are moving to the Chicago exchange ... ?
Ah,....Hop. Nice try.
Growing interest. It will be growing interest in Brainchip that increases volume.
Actually I have no idea what the catalyst will be. I just know the volume will be huge.
Space Cadet
Regular
Thanks mate. Still can't read it but I do need new glasses. Missus tried too but could only get half of it. Will keep an eye on Booz Allen. Interesting stuff.
SC
CHIPS
Regular
Even with better glasses, you won't be able to read it, I guess. I can't read all of it either. I could not get a better picture.
ChatGPT was able to read it
TOP TITLE:
C2BMC
Command and Control, Battle Management, and Communications
Subtitle (below C2BMC):
The Command and Control, Battle Management, and Communications (C2BMC) program is the hub of the layered Missile Defense System. It is a vital operational system that enables the President, Secretary of Defense and Combatant Commanders at strategic, regional and operational levels to systematically plan missile defense operations, to collectively see the battle develop, and to dynamically manage shielded networked sensors and weapons systems to achieve global and regional mission objectives.
TOP COMMAND STRIP (color-coded):
- NMCC
- USSPACECOM
- USNORTHCOM
- USINDOPACOM
- USEUCOM
- USCENTCOM
MAIN TITLE (center of the image):
THE SYSTEM OF ELEMENTS
DEFENSE SEGMENTS (left to right):
BOOST
Defense Segment
(Shows missiles launching)
ASCENT / MIDCOURSE
Defense Segment
(Depicts missile intercept systems)
- GBI
Ground-Based Interceptor - SM-3 IIA
Standard Missile - SM-3 IA/IB
Standard Missile - AEGIS
SHIP & ASHORE
Ballistic Missile Defense
TERMINAL
Defense Segmen
- THAAD
Terminal High Altitude Area Defense - SM-6
Standard Missile - AEGIS
(Standard Missile) - PAC-3
Patriot Advanced Capability
Title:
SENSORS
Descriptive Text:
An effective layered defense relies on timely sensor information provided by space and ground-based assets. Sensors provide critical tracking and discrimination data to help warfighters see the threat, track it and engage it effectively.
Sensor Types (with icons):
- SATELLITE SURVEILLANCE SYSTEMS
- UPGRADED EARLY WARNING RADARS
- FORWARD-BASED RADARS
- SPY RADARS
- DISCRIMINATING RADARS
Fullmoonfever
Top 20
Thesis just out of KTH.
Haven't downloaded to read it but just the abstract was enough for me on the confirmation of how just our AKD1000 First Gen stacked up and clearly showed its "edge" in the edge space over NVIDIA GPU.
Think it's a given that as complexity increases the processing power shifts however the power usage is still an advantage.
Be interesting to see how it goes when they focus on fully customised models.
Would also be great to see what step up AKD1500 / Gen 2 / TENNs etc is capable of in a comparison given this was just Gen 1.
Comparison of Akida Neuromorphic Processor and NVIDIA Graphics Processor Unit for Spiking Neural Networks
2025 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesisAlternative title
Jämförelse av neuromorfisk processor Akida och NVIDIA grafikkort för Spiking Neural Networks (Swedish)
This thesis investigates the latency, throughput and energy efficiency of the BrainChip Akida AKD1000 neuromorphic processor compared to a NVIDIA GeForce GTX 1080 when running two different spiking neural network models on both hardwares. Spiking neural networks is a subset of neural networks that are specialized for neuromorphic processor. The first model is a simple image classification model (GXNOR on MNIST), and the second is a more complex object detection model (YOLOv2 on Pascal VOC). The models were trained and quantized to 2-bit and 4-bit weight precision, respectively, enabling spiking execution both on Akida AKD1000 and on GTX 1080, for the GPU CUDA was used. Results show that Akida achieved significant reductions in energy consumption and clock cycles for both models, consistent with prior findings within the field. Specifically, for the simple classification model the AKD1000 achieved 99.5 % energy reduction with 76.7 % faster inference times, despite having a clock rate 91.5 % slower than the GPU. However, for the more complex object detection model, the Akida took 118.1 % longer per inference, while reducing the energy expenditure by 96.0 %. For the MNIST model the AKD1000 showed no correlation in both cycles & time and cycle & energy. While for the YOLOv2 model it had a 0.2 correlation for both previous mentioned ratios. Suggesting that as model complexity increases, the Akida’s behaviour converges toward the GPU’s linear correlation patterns. In conclusion, the AKD1000 processor demonstrates clear advantages for low-power, edge-oriented applications where latency and efficiency are critical. However, these benefits diminish with increasing model complexity, where GPUs maintain superior scalability and performance. Due to limited documentation of the chosen models, a 1-to-1 comparison was not possible. Future work should focus on fully customized models to further explore the dynamics.
Haven't downloaded to read it but just the abstract was enough for me on the confirmation of how just our AKD1000 First Gen stacked up and clearly showed its "edge" in the edge space over NVIDIA GPU.
Think it's a given that as complexity increases the processing power shifts however the power usage is still an advantage.
Be interesting to see how it goes when they focus on fully customised models.
Would also be great to see what step up AKD1500 / Gen 2 / TENNs etc is capable of in a comparison given this was just Gen 1.
Comparison of Akida Neuromorphic Processor and NVIDIA Graphics Processor Unit for Spiking Neural Networks
Chemnitz, Carl
KTH, School of Electrical Engineering and Computer Science (EECS).Ermis, Malik
KTH, School of Electrical Engineering and Computer Science (EECS).2025 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesisAlternative title
Jämförelse av neuromorfisk processor Akida och NVIDIA grafikkort för Spiking Neural Networks (Swedish)
Abstract [en]
This thesis investigates the latency, throughput and energy efficiency of the BrainChip Akida AKD1000 neuromorphic processor compared to a NVIDIA GeForce GTX 1080 when running two different spiking neural network models on both hardwares. Spiking neural networks is a subset of neural networks that are specialized for neuromorphic processor. The first model is a simple image classification model (GXNOR on MNIST), and the second is a more complex object detection model (YOLOv2 on Pascal VOC). The models were trained and quantized to 2-bit and 4-bit weight precision, respectively, enabling spiking execution both on Akida AKD1000 and on GTX 1080, for the GPU CUDA was used. Results show that Akida achieved significant reductions in energy consumption and clock cycles for both models, consistent with prior findings within the field. Specifically, for the simple classification model the AKD1000 achieved 99.5 % energy reduction with 76.7 % faster inference times, despite having a clock rate 91.5 % slower than the GPU. However, for the more complex object detection model, the Akida took 118.1 % longer per inference, while reducing the energy expenditure by 96.0 %. For the MNIST model the AKD1000 showed no correlation in both cycles & time and cycle & energy. While for the YOLOv2 model it had a 0.2 correlation for both previous mentioned ratios. Suggesting that as model complexity increases, the Akida’s behaviour converges toward the GPU’s linear correlation patterns. In conclusion, the AKD1000 processor demonstrates clear advantages for low-power, edge-oriented applications where latency and efficiency are critical. However, these benefits diminish with increasing model complexity, where GPUs maintain superior scalability and performance. Due to limited documentation of the chosen models, a 1-to-1 comparison was not possible. Future work should focus on fully customized models to further explore the dynamics.
I think AKIDA 1500 is a big step up from the 1000.Thesis just out of KTH.
Haven't downloaded to read it but just the abstract was enough for me on the confirmation of how just our AKD1000 First Gen stacked up and clearly showed its "edge" in the edge space over NVIDIA GPU.
Think it's a given that as complexity increases the processing power shifts however the power usage is still an advantage.
Be interesting to see how it goes when they focus on fully customised models.
Would also be great to see what step up AKD1500 / Gen 2 / TENNs etc is capable of in a comparison given this was just Gen 1.
Comparison of Akida Neuromorphic Processor and NVIDIA Graphics Processor Unit for Spiking Neural Networks
Chemnitz, Carl
KTH, School of Electrical Engineering and Computer Science (EECS).
Ermis, Malik
KTH, School of Electrical Engineering and Computer Science (EECS).
2025 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesisAlternative title
Jämförelse av neuromorfisk processor Akida och NVIDIA grafikkort för Spiking Neural Networks (Swedish)
Abstract [en]
This thesis investigates the latency, throughput and energy efficiency of the BrainChip Akida AKD1000 neuromorphic processor compared to a NVIDIA GeForce GTX 1080 when running two different spiking neural network models on both hardwares. Spiking neural networks is a subset of neural networks that are specialized for neuromorphic processor. The first model is a simple image classification model (GXNOR on MNIST), and the second is a more complex object detection model (YOLOv2 on Pascal VOC). The models were trained and quantized to 2-bit and 4-bit weight precision, respectively, enabling spiking execution both on Akida AKD1000 and on GTX 1080, for the GPU CUDA was used. Results show that Akida achieved significant reductions in energy consumption and clock cycles for both models, consistent with prior findings within the field. Specifically, for the simple classification model the AKD1000 achieved 99.5 % energy reduction with 76.7 % faster inference times, despite having a clock rate 91.5 % slower than the GPU. However, for the more complex object detection model, the Akida took 118.1 % longer per inference, while reducing the energy expenditure by 96.0 %. For the MNIST model the AKD1000 showed no correlation in both cycles & time and cycle & energy. While for the YOLOv2 model it had a 0.2 correlation for both previous mentioned ratios. Suggesting that as model complexity increases, the Akida’s behaviour converges toward the GPU’s linear correlation patterns. In conclusion, the AKD1000 processor demonstrates clear advantages for low-power, edge-oriented applications where latency and efficiency are critical. However, these benefits diminish with increasing model complexity, where GPUs maintain superior scalability and performance. Due to limited documentation of the chosen models, a 1-to-1 comparison was not possible. Future work should focus on fully customized models to further explore the dynamics.
One thing that we learnt from taping out the AKIDA 1000 chip was that minor improvements can be made with little cost to the original taping but major adjustments like the AKIDA1500 has to be completely 'retaped' from scratch which is expensive.
The fast pace of change weighed against taping out costs was likely at least a small factor in not taping out Gen2.
AKIDA 1000 and 1500 gets the chips 'out there'.
I am surprised we have not heard more about engagements utilising the1500.
Bascom Hunter purchased $100k worth.
8 bit processing, improved efficiency and the VIT of the1500 is a big step up IMO.
We are thrilled to announce that emotion3D has entered into a definite agreement to be acquired by indie.inc (Nasdaq: INDI) a global leader in automotive innovation. | emotion3D
We are thrilled to announce that emotion3D has entered into a definite agreement to be acquired by indie.inc (Nasdaq: INDI) a global leader in automotive innovation. The acquisition is subject to regulatory clearance and expected to close in the 4th quarter of this year. This marks a...
Similar threads
