BRN Discussion Ongoing

[smoothsailing18]

If Van neu man system have bottleneck in limited computations, then in memory processing have even bigger bottlenecks and that is limited memory.
How is processing going to evolve beyond a certain point with limited memory?
With neurophonic the memory is not only limited with ultra ultra limited.
Dyor

 

[rgupta]

So you mean software can handle hard ware bottlenecks??
Yes to an extent but to grow, hardware have to be improved as well.
To develop better softwares you need a lot of head counts.
On the other hand if the model sizes can be reduced, that can improve performance of Van neu man architect as well, as proven by deep seek.
Dyor

 

[Labsy]

So you mean software can handle hard ware bottlenecks??
Yes to an extent but to grow, hardware have to be improved as well.
To develop better softwares you need a lot of head counts.
On the other hand if the model sizes can be reduced, that can improve performance of Van neu man architect as well, as proven by deep seek.
Dyor

Look buddy,
I don't think the SP will get any lower. You might as well by back in now rather the risk missing the boat.

 

[manny100]

Neuromorphic Chips: Brain-Inspired AI Is Coming Faster Than You Think | InvestorPlace

"The Final Word on the Silent Revolution That Could Outperform Nvidia​

Neuromorphic computing isn’t just the next chip upgrade; it’s a radical leap forward. These brain-inspired systems promise to make machines smarter, faster, and far more energy-efficient.

If they deliver, they won’t just improve AI… they’ll redefine it.

And like every breakthrough before it, the biggest gains go to those who get in before the crowd catches on.

This is the kind of opportunity that could turn small-cap pioneers into market leaders – and supercharge the incumbents building tomorrow’s AI infrastructure.

It may be early days for neuromorphic computing, but it’s no longer theoretical.

The seeds are planted. The architecture is real. And the use cases are arriving fast.

In fact, there’s one we’re particularly bullish on… humanoid robotics.

Humanoid robotics is no longer sci-fi … it’s the next frontier of AI. These machines require real-time, low-power intelligence that only neuromorphic and next-gen AI chips can deliver. As adoption accelerates, the companies building this tech could anchor a trillion-dollar disruption.

That’s why we’re zeroed in on this space.

Today’s breakthroughs in robotics are possible because AI chips are finally fast enough to simulate robots entirely in virtual space: testing millions of actions digitally before a single move happens in the real world.

Imagine a robot learning to sew by practicing millions of stitches overnight, without ever touching fabric.

This shift is reaching a tipping point, revealing the kind of opportunity that could turn small-cap pioneers into market leaders. And we’re got our sights set on one little-known firm ready to step into the limelight."

 

[HopalongPetrovski]

The lovely 'Anastasi in Tech' presents an interesting 30 minutes on the criticality of eco systems.
Worth a look on a Sunday arvo.

 

[rgupta]

Look buddy,
I don't think the SP will get any lower. You might as well by back in now rather the risk missing the boat.

Thanks Labsy, i actually never left the boat.
Cheers

 

[Labsy]

Thanks Labsy, i actually never left the boat.
Cheers

Why always so pessimistic then? Are you superstitious?

 

[7für7]

Everyone keeps talking about getting on the boat or leaving it…
Meanwhile I’m already sitting alone in the rocket.
No wonder it hasn’t launched yet — it’s still waiting for the hillbillies taking their sweet time paddling a boat to the launch pad.

Come on boys and girls, hurry up!

 

[rgupta]

Why always so pessimistic then? Are you superstitious?

Sorry it is not pessimistic but realistic. To attract more public they use some sophisticated words, but understanding the true meaning for the same will let you know what the actual issue is.
Like van neu man bottleneck, and this word is used thousands of times, that akida is out of that. But if you think what is a the bottleneck and what is the akida bottleneck, you will realise akida have bigger bottle neck than van neuman architect.
Realising your limitations is not pessimistic but a sign of strength. Akida knows that and that is why they are trying with the edge devices, where memory is not required that much. But to compete with van neu architect and GPUs with that big limitation is a big task. That is why brainchip does not consider NVIDIA as a competitor but many posters start comparing akida with GPUs which to an extent is no match.
Yes akida can make GPUs 10 times faster if it can be intergrated into a gpu as event based decision making but still no work or no idea is proposed as yet.
I like akida as an accelerator where the utilisation of technology with existing ones is more beneficial than stand alone.
Just my view. And always happy to be corrected

 

[smoothsailing18]

Sorry it is not pessimistic but realistic. To attract more public they use some sophisticated words, but understanding the true meaning for the same will let you know what the actual issue is.
Like van neu man bottleneck, and this word is used thousands of times, that akida is out of that. But if you think what is a the bottleneck and what is the akida bottleneck, you will realise akida have bigger bottle neck than van neuman architect.
Realising your limitations is not pessimistic but a sign of strength. Akida knows that and that is why they are trying with the edge devices, where memory is not required that much. But to compete with van neu architect and GPUs with that big limitation is a big task. That is why brainchip does not consider NVIDIA as a competitor but many posters start comparing akida with GPUs which to an extent is no match.
Yes akida can make GPUs 10 times faster if it can be intergrated into a gpu as event based decision making but still no work or no idea is proposed as yet.
I like akida as an accelerator where the utilisation of technology with existing ones is more beneficial than stand alone.
Just my view. And always happy to be corrected

Nun of that made any sence

 

[FJ-215]

Nun of that made any sence

Please tell me that was deliberate?

 

[7für7]

Sorry it is not pessimistic but realistic. To attract more public they use some sophisticated words, but understanding the true meaning for the same will let you know what the actual issue is.
Like van neu man bottleneck, and this word is used thousands of times, that akida is out of that. But if you think what is a the bottleneck and what is the akida bottleneck, you will realise akida have bigger bottle neck than van neuman architect.
Realising your limitations is not pessimistic but a sign of strength. Akida knows that and that is why they are trying with the edge devices, where memory is not required that much. But to compete with van neu architect and GPUs with that big limitation is a big task. That is why brainchip does not consider NVIDIA as a competitor but many posters start comparing akida with GPUs which to an extent is no match.
Yes akida can make GPUs 10 times faster if it can be intergrated into a gpu as event based decision making but still no work or no idea is proposed as yet.
I like akida as an accelerator where the utilisation of technology with existing ones is more beneficial than stand alone.
Just my view. And always happy to be corrected

Since I don’t want to interfere directly as a neutral observer, I asked ChatGPT for its view — just as a reference point regarding correction :

1. Von Neumann Bottleneck vs. Akida Bottleneck

He claims Akida has an even bigger bottleneck than the classical von Neumann architecture. The fact is: the opposite is true. Reducing these bottlenecks is the very core of Akida’s approach. Akida processes data locally (in-memory computing, spiking), while von Neumann architectures suffer from the constant shuttling of data between CPU and memory.

2. “Memory isn’t so important at the edge”

Here he mixes things up. Yes, edge applications often don’t need huge memory banks like GPUs. But saying Akida is therefore only limited to edge use is short-sighted. Akida scales: from sensors/edge up to larger embedded applications.

3. “BrainChip doesn’t compare itself with NVIDIA”

True, BrainChip doesn’t position Akida as a GPU replacement. But Akida complements GPUs/CPUs (as a co-processor/accelerator). That’s exactly what makes the technology exciting → hybrid solutions that make GPUs faster, more efficient, and more power-saving.


4. “No ideas have been proposed yet”

That’s simply wrong. BrainChip has repeatedly shown how Akida can be integrated: with Edge Impulse for sensor applications, with prototyping boards (M.2, PCIe) for developers, and through partners like Renesas or MegaChips, enabling integration into SoCs.


5. “Akida as an accelerator makes sense”

This is the only point he gets right. But he frames it as if it were a limitation. In reality, it’s precisely the business model: Akida is sold/licensed as an accelerator/IP block, not as a standalone mega-chip.

 

[Tothemoon24]

Today I came across a post about the latest updates on Akida, and it reminded me of a project I had the chance to work on a few months ago, back when it wasn’t on many people’s radar yet, and most importantly, when we worked directly with the physical processor.

At the time, my manager shared this post highlighting the work we did, and it makes me proud to see how relevant that experience still is now that the technology is gaining traction.

Now that Akida is back in the spotlight, I’m excited to discover what's new. Happy to connect!

 

[Tothemoon24]