BRN Discussion Ongoing

[Slade]

Happy to get a response from Tony Lewis and it gives me confidence that we are making an impact at trade fairs. Thanks also to the members that shared their experiences of trade fairs.

 

[Guzzi62]

Happy to get a response from Tony Lewis and it gives me confidence that we are making an impact at trade fairs. Thanks also to the members that shared their experiences of trade fairs.

View attachment 79373

On LinkedIn?

But nice nevertheless.

 

[Rach2512]

Alf Kuchenbuch on LinkedIn: #embeddedworld #risc_v

Day 3 of #EmbeddedWorld in Nürnberg, Germany. Come and see our Akida 2.0 in action at our booth 213 in Hall 5, next to the #RISC_V booth!

www.linkedin.com

Nice comment from Spencer Huang of Edge Impulse.

 

[Slade]

On LinkedIn?

But nice nevertheless.

Yes on LinkedIn. Sorry I couldn’t figure out how to link it, no pun intended.

 

[7für7]

Alf Kuchenbuch on LinkedIn: #embeddedworld #risc_v

Day 3 of #EmbeddedWorld in Nürnberg, Germany. Come and see our Akida 2.0 in action at our booth 213 in Hall 5, next to the #RISC_V booth!

www.linkedin.com

Nice comment from Spencer Huang of Edge Impulse.

“Nice?” is just a plain, generic comment. Or maybe I’m just so jaded that I’m missing the message.. could you give me a hint?

 

[MadMayHam | 合氣道]

Alf Kuchenbuch on LinkedIn: Embedded World 2025, what's cooking at BrainChip : Our CTO M Anthony Lewis…

Embedded World 2025, what's cooking at BrainChip : Our CTO M Anthony Lewis and the BrainChip team present our demos, straight from the lab: Akida 2.0 IP…

www.linkedin.com

View attachment 79180

Thanks for the link Supersonic. The irony of his presentation is that in his Linkedin post, Alf Kuchenbuch commented "Please excuse the poor sound quality of this clip, we took it at the show floor." while Tony was talking about, AND PRESENTING an AN AUDIO DENOISING solution that was denoising THE VERY SAME SHOWFLOOR COMMOTION.

Another missed opportunity for the company -- they could have started the clip with the ambient sound, and then denoised the rest of the clip to show its efficacy.

 

[Guzzi62]

Thanks for the link Supersonic. The irony of his presentation is that in his Linkedin post, Alf Kuchenbuch commented "Please excuse the poor sound quality of this clip, we took it at the show floor." while Tony was talking about, AND PRESENTING an audio denoising solution that was denoising THE VERY SAME SHOWFLOOR COMMOTION.

Another missed opportunity for the company -- they could have started the clip with the ambient sound, and then they should have denoised the rest of the clip to show its efficacy.

Nope, it was fine, people in the know got the message.

Keep on with your negative crap, one day you might succeed finding something that sticks.

 

[Rach2512]

“Nice?” is just a plain, generic comment. Or maybe I’m just so jaded that I’m missing the message.. could you give me a hint?

They are all still friends even though Edge Impulse are being bought out by Qualcom. I took it as positive. Does anyone else agree or have I got it wrong

 

[7für7]

They are all still friends even though Edge Impulse are being bought out by Qualcom. I took it as positive. Does anyone else agree or have I got it wrong

Why should they not be friends? If they was bought out doesn’t mean they’re not talking together anymore.. they keep in touch it’s normal. I don’t mean it negative but I don’t think I will interpreting too much about it

 

[Diogenese]

They are all still friends even though Edge Impulse are being bought out by Qualcom. I took it as positive. Does anyone else agree or have I got it wrong

My reading is the Qualcomm/EI union is an open marriage.

They already have a lot of mutual friends. Perhaps it may involve a bit of partner-swapping, but that will involve more than the keys in a bowl, and more of the "You show me yours, ..." .

 

[MadMayHam | 合氣道]

Nope, it was fine, people in the know got the message.

Keep on with your negative crap, one day you might succeed finding something that sticks.

I'm glad you've got your blinkers on and live in your dreamland, and see only what you want to see, but I've only stated facts.

Learn to read the room... the overall sentiment here isn't positive toward the co's management.

 

[GStocks123]

More relating to the Apple job advert I recently posted

Apple M4 Chip and the Future of Apple Silicon (Apple M5) - LearnXYZ

Apple’s transition from Intel processors to custom-designed Apple Silicon marked…

learnxyz.in

Anticipated Innovations for the Apple M5 Chipset – The Next Frontier in Silicon​

Following the groundbreaking M4 chipset, Apple’s M5 is poised to push the boundaries of performance, efficiency, and artificial intelligence even further. While Apple has not officially confirmed details, industry trends, patent filings, and TSMC’s roadmap provide a compelling glimpse into what the M5 might bring. Here’s a speculative deep dive into its potential advancements:

---

1. Manufacturing Process: TSMC’s 1.8nm (A14) Node

• Shrinking to 1.8nm: The M5 could adopt TSMC’s next-generation 1.8nm (A14) process by late 2026, offering:
  • ~15% higher transistor density vs. 2nm (M4), enabling 50+ billion transistors.
  • 20-25% power efficiency gains and 10-15% faster clock speeds at matched power.
  • Backside Power Delivery (BSPDN): A new technology that moves power rails beneath the silicon, reducing signal interference and improving performance for high-frequency designs.
• 3D Hybrid Bonding: Stacked cache or memory layers using TSMC’s 3D Fabric™ for ultra-low latency between CPU, GPU, and Neural Engine.

2. CPU and GPU: Performance Redefined

A. CPU Architecture:

• Next-Gen “Hurricane” P-Cores:
  • 12-16 performance cores with SMT (Simultaneous Multi-Threading) support, enabling 24-32 threads for workstation-grade multitasking.
  • Clock speeds exceeding 4.5 GHz (up from M4’s ~4.2 GHz).
• “Tempest” E-Cores:
  • Redesigned efficiency cores with 30% faster single-threaded performance than M4’s E-cores, closing the gap with Intel’s P-cores.

B. GPU Evolution:

• Ray Tracing Pro Max:
  • Dedicated hardware for path tracing (cinema-quality lighting) and AI-upscaled frame generation (similar to NVIDIA DLSS 3).
  • Up to 64 GPU cores in M5 Ultra variants, targeting AAA gaming and 8K/120fps video workflows.
• Apple Photon Architecture:
  • A rumored in-house GPU design replacing ARM’s Immortalis, with support for DirectX 12 Ultimate and OpenGL Next.

3. AI and Neural Engine: The Brain of the Future

• 6th-Gen Neural Engine:
  • 128-core design (up from M4’s 24 cores), delivering 200+ TOPS (trillion operations per second).
  • Transformer Accelerators: Dedicated hardware for real-time inference of 100B+ parameter LLMs (e.g., ChatGPT-scale models on-device).
• Autonomous AI Agents:
  • System-level integration for AI agents that automate workflows (e.g., drafting emails, editing videos) without cloud dependency.
• Neuromorphic Computing:
  
  • Experimental support for spiking neural networks (SNNs), mimicking human brain efficiency for sensor data processing in Vision Pro

 

[Diogenese]

More relating to the Apple job advert I recently posted

Apple M4 Chip and the Future of Apple Silicon (Apple M5) - LearnXYZ

Apple’s transition from Intel processors to custom-designed Apple Silicon marked…

learnxyz.in

Anticipated Innovations for the Apple M5 Chipset – The Next Frontier in Silicon​

Following the groundbreaking M4 chipset, Apple’s M5 is poised to push the boundaries of performance, efficiency, and artificial intelligence even further. While Apple has not officially confirmed details, industry trends, patent filings, and TSMC’s roadmap provide a compelling glimpse into what the M5 might bring. Here’s a speculative deep dive into its potential advancements:

---

1. Manufacturing Process: TSMC’s 1.8nm (A14) Node

• Shrinking to 1.8nm: The M5 could adopt TSMC’s next-generation 1.8nm (A14) process by late 2026, offering:
  • ~15% higher transistor density vs. 2nm (M4), enabling 50+ billion transistors.
  • 20-25% power efficiency gains and 10-15% faster clock speeds at matched power.
  • Backside Power Delivery (BSPDN): A new technology that moves power rails beneath the silicon, reducing signal interference and improving performance for high-frequency designs.
• 3D Hybrid Bonding: Stacked cache or memory layers using TSMC’s 3D Fabric™ for ultra-low latency between CPU, GPU, and Neural Engine.

2. CPU and GPU: Performance Redefined

A. CPU Architecture:

• Next-Gen “Hurricane” P-Cores:
  • 12-16 performance cores with SMT (Simultaneous Multi-Threading) support, enabling 24-32 threads for workstation-grade multitasking.
  • Clock speeds exceeding 4.5 GHz (up from M4’s ~4.2 GHz).
• “Tempest” E-Cores:
  • Redesigned efficiency cores with 30% faster single-threaded performance than M4’s E-cores, closing the gap with Intel’s P-cores.

B. GPU Evolution:

• Ray Tracing Pro Max:
  • Dedicated hardware for path tracing (cinema-quality lighting) and AI-upscaled frame generation (similar to NVIDIA DLSS 3).
  • Up to 64 GPU cores in M5 Ultra variants, targeting AAA gaming and 8K/120fps video workflows.
• Apple Photon Architecture:
  • A rumored in-house GPU design replacing ARM’s Immortalis, with support for DirectX 12 Ultimate and OpenGL Next.

3. AI and Neural Engine: The Brain of the Future

• 6th-Gen Neural Engine:
  • 128-core design (up from M4’s 24 cores), delivering 200+ TOPS (trillion operations per second).
  • Transformer Accelerators: Dedicated hardware for real-time inference of 100B+ parameter LLMs (e.g., ChatGPT-scale models on-device).
• Autonomous AI Agents:
  • System-level integration for AI agents that automate workflows (e.g., drafting emails, editing videos) without cloud dependency.
• Neuromorphic Computing:
  
  • Experimental support for spiking neural networks (SNNs), mimicking human brain efficiency for sensor data processing in Vision Pro

I think TENNS are a step beyond Transformers, but I don't have a reference to hand.

 

[Diogenese]

I think TENNS are a step beyond Transformers, but I don't have a reference to hand.

Found it!

https://www.wevolver.com/article/2024-state-of-edge-ai-report/generative-ai-at-the-edge

"Combining the new algorithms with some innovative hardware choices – especially those that make 3D convolutions very efficient – brings a big step in efficiency—demonstrating more than 100-500x improvement in energy efficiency without compromising accuracy. In terms of disruptive potential, TENNs could revolutionize LLMs and LVMs at the Edge. Tested on prior generation transformer-based models, a TENNs-equivalent has shown that for equivalent perplexity scores (an indicator of correctness), while reducing model size and MACs/token by 3-4 orders of magnitude. More importantly, the training of these models is similar to that of CNN training and yet takes less than 1/10th the time compared to the transformer equivalent."

 

[Diogenese]

@Bravo

MB CLA 792 km
https://www.linkedin.com/feed/update/urn:li:activity:7306011210758381572/

 

[7für7]

@Bravo WHAT THE HECK… I KNEW IT!!!

 

[Pom down under]

Dont think this short video has been posted?

 

[Pom down under]

Gesture Appliances Control with Pose Detection - BrainChip AKD1000 | Expert Projects

Control your TV, Air Conditioner or Lightbulb by just pointing your finger at them, using the BrainChip AKD1000 achieving great accuracy and low power consumption.

docs.edgeimpulse.com

 

[Adam82]

We are the architects of our new Mercedes-Benz Operating System MB.OS. It… | Markus Schäfer | 25 comments

We are the architects of our new Mercedes-Benz Operating System MB.OS. It enables us to take the customer experience to the next level, making full use of… | 25 comments on LinkedIn

www.linkedin.com

 

[Deadpool]

My reading is the Qualcomm/EI union is an open marriage.

They already have a lot of mutual friends. Perhaps it may involve a bit of partner-swapping, but that will involve more than the keys in a bowl, and more of the "You show me yours, ..." .

Can only hope they have a good supply of antibiotics