BRN Discussion Ongoing
- Thread starter TechGirl
- Start date
smoothsailing18
Regular
From the other site
Here are some benchmarks comparing PointNet++ running on Akida 2 vs NVIDIA Jetson Xavier NX and Orin NX for real-time LiDAR classification at the edge.
*GPT5
Here are some benchmarks comparing PointNet++ running on Akida 2 vs NVIDIA Jetson Xavier NX and Orin NX for real-time LiDAR classification at the edge.
1. Benchmark Sources
- Akida PointNet++: BrainChip LiDAR Point Cloud Model brochure (Oct 2025).
- Jetson Xavier NX / Orin NX: Derived from public PyTorch PointNet++ benchmarks on ModelNet40 and KITTI, using TensorRT-optimized inference (batch = 1).
- Xavier NX: 384 CUDA cores, 21 TOPS INT8
- Orin NX: 1024 CUDA cores, 100 TOPS INT8
- Power figures are measured in NVIDIA’s MaxN mode, which reflects real deployment on drones / robotics platforms.
2. PointNet++ Performance Comparison
| Metric | Akida 2 | Jetson Xavier NX | Jetson Orin NX |
|---|---|---|---|
| FPS (ModelNet40) | 183 FPS | 65 – 85 FPS (FP16/INT8) | 110 – 135 FPS (FP16/INT8) |
| Latency / Frame | 5 – 6 ms | 12 – 15 ms | 7 – 9 ms |
| Power | 50 mW | 10 – 15 W | 15 – 25 W |
| Energy / Inference | 0.28 mJ | ~150 – 200 mJ | ~200 – 300 mJ |
| Model Size | 1.07 MB | ~10 – 12 MB | ~10 – 12 MB |
| Accuracy (ModelNet40) | 81.6 % (4-bit QAT) | 89 – 90 % (FP32 baseline) | 89 – 90 % (FP32 baseline) |
| Deployment Mode | Always-on, ultra-low power | Embedded GPU (fan/heat dissipation required) | High-end embedded GPU |
3. What This Shows
- Throughput
- Akida 2 actually outperforms Xavier and Orin NX on raw FPS, despite being neuromorphic and consuming three orders of magnitude less power.
- Orin NX gets closer but still lags slightly at similar batch sizes.
- Power & Energy
- Akida’s ~50 mW is in a completely different regime than the 10–25 W Jetson modules.
- That’s ~500×–1000× lower energy per inference, which is decisive for always-on payloads (e.g. drones, satellites, smart infrastructure).
- Accuracy Trade-off
- Akida’s quantized model (4-bit QAT) loses ~8 % absolute accuracy vs FP32, but this is expected and often acceptable for edge classification, especially if upstream sensor fusion provides redundancy.
- Form Factor / Thermal
- Jetsons need active cooling and steady power supply — not trivial on space or micro-UAV platforms.
- Akida can operate fanless, battery-powered, or solar-powered.
4. Strategic Takeaway
| Use Case | Best Fit |
|---|---|
| Battery-powered always-on LiDAR classification (e.g., satellite autonomy, drones, infrastructure nodes) | Akida 2 — ultra-low power, high FPS, compact |
| Onboard AI co-processor with larger perception stack (e.g., autonomous cars, ground robots) | Jetson Orin NX — higher model flexibility, better FP32 accuracy, but power-hungry |
| Mixed sensor fusion payloads with strict SWaP (e.g., ESA cubesats, tactical drones) | Akida as front-end classifier + Jetson/FPGA for downstream fusion or planning |
Summary Table
| Feature | Akida 2 | Jetson Xavier NX | Jetson Orin NX |
|---|---|---|---|
| FPS | 183 | 65–85 | 110–135 |
| Power | 0.05 W | 10–15 W | 15–25 W |
| Energy/Inference | 0.28 mJ | 150–200 mJ | 200–300 mJ |
| Accuracy | 81.6 % | ≈ 90 % | ≈ 90 % |
| Edge Suitability | Always-on | Thermally constrained | High-end only |
Bottom line:
For PointNet++ at the edge, Akida 2 outperforms Jetson Xavier NX and Orin NX on raw FPS, power, and energy efficiency, with a modest accuracy gap from quantization that can be narrowed through improved training and model updates. This is why BrainChip is targeting Akida PointNet++ for autonomous drones, satellites, and infrastructure nodes — it's built for tiny, always-on LiDAR intelligence rather than general AI workloads.*GPT5
Tothemoon24
Top 20
Stocks are rising & so am I 
The Hidden Tech in Your Next Wearable: Why Spiking is the New Speed and Efficiency in Hardware.
Most of the AI buzz focuses on massive models in cloud data centers. But the real revolution in low-power intelligence is happening at the very edge: through Neuromorphic Computing.
The challenge for devices like advanced health trackers, industrial sensors, or autonomous drone navigation is simple: continuous processing with crippling power constraints. Running traditional Deep Learning on these devices drains the battery almost instantly.
The solution is brilliant—and brain-inspired.
Enter the Spiking Neural Network (SNN):
Unlike a conventional chip that processes data in large, power-hungry blocks, neuromorphic chips use Spiking Neural Networks (SNNs). These networks only fire (process) a signal when the input data exceeds a certain threshold—just like a biological neuron. This is known as event-based processing.
Why this is a game-changer for niche industries:
1. Extreme Efficiency: A 100x to 1000x improvement in energy efficiency compared to standard GPUs for certain tasks. Imagine a sensor that lasts for years, not days.
2. Instant Reaction: Processing occurs in real-time, right where the data is created (Edge AI), making it perfect for safety-critical systems like autonomous vehicles or real-time medical monitoring where latency can't be tolerated.
3. Adaptive Learning: The SNN architecture inherently supports "one-shot" or continuous on-device learning, adapting to its environment without massive retraining cycles.
If you are a Product Manager, Hardware Engineer, or Investor building solutions that require high-speed, continuous processing under severe power budgets, you need to be integrating this ecosystem now.
What niche application (outside of health) is perfectly primed for a move to neuromorphic processing? Share your 'killer app' idea below!
#NeuromorphicComputing #EdgeAI #HardwareInnovation #IoT #SNN #FutureofTech #DeepLearning
www.linkedin.com
The Hidden Tech in Your Next Wearable: Why Spiking is the New Speed and Efficiency in Hardware.
Most of the AI buzz focuses on massive models in cloud data centers. But the real revolution in low-power intelligence is happening at the very edge: through Neuromorphic Computing.
The challenge for devices like advanced health trackers, industrial sensors, or autonomous drone navigation is simple: continuous processing with crippling power constraints. Running traditional Deep Learning on these devices drains the battery almost instantly.
The solution is brilliant—and brain-inspired.
Enter the Spiking Neural Network (SNN):
Unlike a conventional chip that processes data in large, power-hungry blocks, neuromorphic chips use Spiking Neural Networks (SNNs). These networks only fire (process) a signal when the input data exceeds a certain threshold—just like a biological neuron. This is known as event-based processing.
Why this is a game-changer for niche industries:
1. Extreme Efficiency: A 100x to 1000x improvement in energy efficiency compared to standard GPUs for certain tasks. Imagine a sensor that lasts for years, not days.
2. Instant Reaction: Processing occurs in real-time, right where the data is created (Edge AI), making it perfect for safety-critical systems like autonomous vehicles or real-time medical monitoring where latency can't be tolerated.
3. Adaptive Learning: The SNN architecture inherently supports "one-shot" or continuous on-device learning, adapting to its environment without massive retraining cycles.
If you are a Product Manager, Hardware Engineer, or Investor building solutions that require high-speed, continuous processing under severe power budgets, you need to be integrating this ecosystem now.
What niche application (outside of health) is perfectly primed for a move to neuromorphic processing? Share your 'killer app' idea below!
#NeuromorphicComputing #EdgeAI #HardwareInnovation #IoT #SNN #FutureofTech #DeepLearning
#neuromorphiccomputing #edgeai #hardwareinnovation #iot #snn #futureoftech #deeplearning | Noushad Yashan ( 牙山 / 야산 / ياشان )
The Hidden Tech in Your Next Wearable: Why Spiking is the New Speed and Efficiency in Hardware. Most of the AI buzz focuses on massive models in cloud data centers. But the real revolution in low-power intelligence is happening at the very edge: through Neuromorphic Computing. The challenge...
smoothsailing18
Regular
FF
AND NOW FROM THE
“VALEO BRAIN DIVISION”
https://www.valeo.com/en/brain-division/
“The emergence of software-defined vehicles (SDV) marks a significant paradigm shift that is reshaping traditional automotive frameworks. Our Valeo Brain Division is actively engaged in this transformation. By accelerating in driving automation with Advanced Driver Assistance Systems (ADAS) and reinventing the interior experience, we are leveraging opportunities offered by the new electric/electronic architectures inherent to the SDV.
Our solutions enhance the safety and enjoyment of driving by using a broad spectrum of sensor technologies (ultrasonic sensors, cameras, thermal imaging, and medium to long-range radars, as well as LiDAR), computing units acting like “brains” (ranging from Zone Controllers and Domain Controllers to Central Computing Units), as well as the software itself with advanced algorithms and artificial intelligence (AI).
To reinvent the interior experience, we develop technology in the field of human-machine interface and vehicle cabin monitoring that improves safety, and that also creates a comfortable cocoon-like interior, offering passengers a personalized, interactive and secure experience.
Lastly, connectivity is key in the SDV transformation. Valeo Brain Division’s solutions offer the high-speed, low-latency connectivity required for state-of-the-art driving automation and interior experience.”
Quite an interesting name given to this new to me Division at Valeo
AND NOW FROM THE
“VALEO BRAIN DIVISION”
https://www.valeo.com/en/brain-division/
“The emergence of software-defined vehicles (SDV) marks a significant paradigm shift that is reshaping traditional automotive frameworks. Our Valeo Brain Division is actively engaged in this transformation. By accelerating in driving automation with Advanced Driver Assistance Systems (ADAS) and reinventing the interior experience, we are leveraging opportunities offered by the new electric/electronic architectures inherent to the SDV.
Our solutions enhance the safety and enjoyment of driving by using a broad spectrum of sensor technologies (ultrasonic sensors, cameras, thermal imaging, and medium to long-range radars, as well as LiDAR), computing units acting like “brains” (ranging from Zone Controllers and Domain Controllers to Central Computing Units), as well as the software itself with advanced algorithms and artificial intelligence (AI).
To reinvent the interior experience, we develop technology in the field of human-machine interface and vehicle cabin monitoring that improves safety, and that also creates a comfortable cocoon-like interior, offering passengers a personalized, interactive and secure experience.
Lastly, connectivity is key in the SDV transformation. Valeo Brain Division’s solutions offer the high-speed, low-latency connectivity required for state-of-the-art driving automation and interior experience.”
Quite an interesting name given to this new to me Division at Valeo
Frangipani
Top 20
Extremely interestingby Fergus Longbottom, who works as a software engineer in Space Domain Awareness / Remote Sensing for Canberra-based Electro Optic Systems (EOS):
View attachment 87642 View attachment 87643
LinkedIn Login, Sign in | LinkedIn
Login to LinkedIn to keep in touch with people you know, share ideas, and build your career.www.linkedin.com
View attachment 87644
View attachment 87645
World-leading defence and space systems technology | EOS
Creating world-leading space and defence systems technology to enhance the security of Australia and its allies.eos-aus.com
View attachment 87646 View attachment 87647
Space Domain Awareness | Electro Optic Systems
EOS has been providing Space Domain Awareness (SDA) services covering low-Earth orbit (LEO), medium-Earth orbit (MEO) and more
View attachment 87655
View attachment 87653 View attachment 87652
EOS also have subsidiaries in the US (https://www.eosdsusa.com/) and NZ (https://kiwistaroptics.com/).
And as you can see, it is not the first time Fergus Longbottom has liked BrainChip or BrainChip-related posts… Which of course doesn’t necessarily mean that he is working with Akida at EOS, but it’s definitely good to know someone in the company is very much aware of us, if not more…
#BrainChipDomainAwareness
LinkedIn Login, Sign in | LinkedIn
Login to LinkedIn to keep in touch with people you know, share ideas, and build your career.
View attachment 87657
View attachment 87658
View attachment 87659
View attachment 87660
View attachment 87661
View attachment 87662
I noticed our company’s latest LinkedIn post about Akida PointNet++ got yet another vote of approval (this time even a
) from Fergus Longbottom, who works as a Software Engineer in Space Domain Awareness/Remote Sensing for Canberra-headquartered Electro Optic Systems (EOS).(Note that there seems to be a timeout issue with their homepage https://eos-aus.com at the moment, but in my two earlier posts tagged above you can see screenshots of that website taken in June).
LinkedIn Login, Sign in | LinkedIn
Login to LinkedIn to keep in touch with people you know, share ideas, and build your career.
Frangipani
Top 20
I don’t recall ever having come across this Akida IP webpage before, although the Akida 1 IP and Akida 2 IP Brochures are not exactly brandnew and were apparently uploaded sometime in April, shortly before the launch of the redesigned BRN website.
Using the
search function here on TSE didn’t yield any results for me either, but then again I kept on getting the message “The following words were not included in your search because they are too short, too long, or too common…”, so it’s a bit hard to tell whether or not this has been posted before.
A good overview in any case…
brainchip.com
AkidaTM is a neural processor platform inspired by the brain’s cognitive capabilities and energy efficiency. It delivers low-power, real-time AI processing at the edge using neuromorphic principles for applications like vision, audio, and sensor fusion.
Akida 1
Cost-effective, predictable design and implementation
Minimizes compute and communication – Minimizes host CPU usage
One-shot/few-shot learning. Minimizes sensitive data sent. Improves security and privacy
Extremely configurable and post-silicon flexibility
Using the
search function here on TSE didn’t yield any results for me either, but then again I kept on getting the message “The following words were not included in your search because they are too short, too long, or too common…”, so it’s a bit hard to tell whether or not this has been posted before.A good overview in any case…
IP
AkidaTM is a neural processor platform inspired by the brain’s cognitive capabilities and energy efficiency. It delivers low-power, real-time AI processing at the edge using neuromorphic principles for applications like vision, audio, and sensor fusion. Self-Contained Neural Processor...
brainchip.com
AkidaTM is a neural processor platform inspired by the brain’s cognitive capabilities and energy efficiency. It delivers low-power, real-time AI processing at the edge using neuromorphic principles for applications like vision, audio, and sensor fusion.
Self-Contained Neural Processor
- Scalable fabric of 1-128 nodes
- Each neural node supports 128MACs
- Configurable 50-130K embedded local SRAM per node
- DMA for all memory and model operations
- Multi-layer execution without host CPU
- Integrate with any Microcontroller or Application Processor
- Efficient algorithmic mesh
Key Features
Akida 1
- Supports 4-, 2-, and 1-bit weights and activations
- Supports multiple layers simultaneously
- Convolutional Neural Processor (CNP) and
- Fully-connected Neural Processor (FNP)
Akida 2
- Supports 8-, 4-, and 1-bit weights and activations
- Programmable Activation Functions
- Skip Connections
- Support for Spatio-Temporal and Temporal Event-Based Neural Network
Silicon-Proven, Fully Digital Neuromorphic Implementation
Cost-effective, predictable design and implementation
Event-Based Hardware Acceleration
Minimizes compute and communication – Minimizes host CPU usage
On-Chip Learning
One-shot/few-shot learning. Minimizes sensitive data sent. Improves security and privacy
Configurable and Scalable
Extremely configurable and post-silicon flexibility
Space Cadet
Regular
Space Cadet
Regular
Space Cadet
Regular
Tothemoon24
Top 20
🚨 BREAKING: SoftBank is making a $5.4B bet on industrial robotics by acquiring ABB Robotics. The company announced it will acquire ABB Robotics (the robot division of ABB), abandoning ABB’s… | Lukas M. Ziegler
🚨 BREAKING: SoftBank is making a $5.4B bet on industrial robotics by acquiring ABB Robotics. The company announced it will acquire ABB Robotics (the robot division of ABB), abandoning ABB’s earlier plan to spin the unit off as a separate listing. The deal is expected to close in mid-to-late...
View attachment 91828 View attachment 91829
🚨 BREAKING: SoftBank is making a $5.4B bet on industrial robotics by acquiring ABB Robotics. The company announced it will acquire ABB Robotics (the robot division of ABB), abandoning ABB’s… | Lukas M. Ziegler
🚨 BREAKING: SoftBank is making a $5.4B bet on industrial robotics by acquiring ABB Robotics. The company announced it will acquire ABB Robotics (the robot division of ABB), abandoning ABB’s earlier plan to spin the unit off as a separate listing. The deal is expected to close in mid-to-late...
Your post is misleading but who am I to judge…
Btw what do you guys think about drone shields new AI software?
www.tipranks.com
DroneShield Unveils Major AI Software Update to Boost Counter-Drone Capabilities - TipRanks.com
DroneShield Limited ( ($AU:DRO) ) has provided an update. DroneShield Limited has announced a significant software release that enhances its counter-drone capabilit...
Mt09
Regular
Delete your post lol, why add the brainchip robot?…View attachment 91828 View attachment 91829🚨 BREAKING: SoftBank is making a $5.4B bet on industrial robotics by acquiring ABB Robotics. The company announced it will acquire ABB Robotics (the robot division of ABB), abandoning ABB’s… | Lukas M. Ziegler
🚨 BREAKING: SoftBank is making a $5.4B bet on industrial robotics by acquiring ABB Robotics. The company announced it will acquire ABB Robotics (the robot division of ABB), abandoning ABB’s earlier plan to spin the unit off as a separate listing. The deal is expected to close in mid-to-late...
Jumpchooks
Regular
I'm keen to see DRO do well as I am an Investor in them also.Btw what do you guys think about drone shields new AI software?
DroneShield Unveils Major AI Software Update to Boost Counter-Drone Capabilities - TipRanks.com
DroneShield Limited ( ($AU:DRO) ) has provided an update. DroneShield Limited has announced a significant software release that enhances its counter-drone capabilit...
None of that matters unless you are a daytrader. Most of us are longs. We need news with revenue, anything else dosent really matter
FJ-215
Regular
If you say so....
I mean, it's kinda weird that this SP jump has happened around a non country wide public holiday and there seems to have been a delay in reporting of the short positions. Oh....and we suddenly have 103 million shares traded in a day ......
Hmmm.....
My bad I guess....
Nothing to see here....
Frangipani
Top 20
Any Akidaholics meeting the below eligibility criteria interested in NASA’s Beyond the Algorithm Challenge that was launched today?
View attachment 79697
NASA Beyond the Algorithm Challenge
NASA Beyond the Algorithm Challenge: Novel Computing Architectures for Flood Analysis The NASA Earth Science Technology Office (ESTO) seeks solutions to complex Earth Science problems using transformative or unconventional computing technologies such as quantum computing, quantum machine...www.nasa-beyond-challenge.org
View attachment 79698 View attachment 79699 View attachment 79706 View attachment 79701
NASA-Beyond-Challenge.org
View attachment 90564
One of the nine finalist teams of NASA’s Beyond the Algorithm: Novel Computing Architectures for Flood Analysis Challenge, consisting of four Columbia University Computer Science students, submitted a solution they named NEO-FLOOD:
“This paper introduces NEO-FLOOD (Neuromorphic Onboard Flood-mapping), a satellite architecture that eliminates this latency by deploying autonomous AI directly in orbit. NEO-FLOOD integrates space-validated neuromorphic processors (Intel Loihi 2, BrainChip Akida) consuming just 2-5W with our novel Spike2Former-Flood algorithm-a spiking neural network optimized for real-time optical SAR fusion onboard small satellites.”
NASA-Beyond-Challenge.org
View attachment 90565
Five weeks ago, I posted about a quartet of Columbia computer science students who had submitted a proposal named NEO-FLOOD for NASA’s Beyond the Algorithm Challenge (launched in March
).“This paper introduces NEO-FLOOD (Neuromorphic Onboard Flood-mapping), a satellite architecture that eliminates this latency by deploying autonomous AI directly in orbit. NEO-FLOOD integrates space-validated neuromorphic processors (Intel Loihi 2, BrainChip Akida) consuming just 2-5W with our novel Spike2Former-Flood algorithm-a spiking neural network optimized for real-time optical SAR fusion onboard small satellites.”
The four young researchers had been picked as one of nine finalist teams who competed against each other at a live pitch event on 17 September, where they presented their ideas to a panel of judges, NASA experts and venture capitalists.
Three winners were chosen, each team being awarded a $100,000 prize!
Among them the NEO-FLOOD team that also won the Audience Favourite Award!
Now it is on to Phase 3:
“In Phase Three, winners will be invited to attend a "Funding 101" webinar course. Additionally, winners will be contacted 12 months after Pitch Event completion for follow up surveys on further challenge research development and implementation.”
NASA Beyond the Algorithm Challenge
NASA Beyond the Algorithm Challenge: Novel Computing Architectures for Flood Analysis The NASA Earth Science Technology Office (ESTO) seeks solutions to complex Earth Science problems using transformative or unconventional computing technologies such as quantum computing, quantum machine...
www.nasa-beyond-challenge.org
Honored that NEO-FLOOD won the NASA Beyond the Algorithm challenge alongside so many other brilliant teams. It was great to work with my teammates Arav Dhoot, Farhaan Siddiqui, and Bhavyansh… | Rohan Timmaraju | 15 comments
Honored that NEO-FLOOD won the NASA Beyond the Algorithm challenge alongside so many other brilliant teams. It was great to work with my teammates Arav Dhoot, Farhaan Siddiqui, and Bhavyansh Sabharwal. Huge thanks to Blue Clarity and NASA - National Aeronautics and Space Administration for an...
Similar threads
