BRN Discussion Ongoing

[Fullmoonfever]

In every presentation on this issue it has never been suggested that the design would need to be changed to take AKIDA smaller than 28nm.

A quick Google brought up this paper which appears to support the proposition that Scratch pad memory can at least move down to 7nm:

“A 7-nm Compute-in-Memory SRAM Macro Supporting Multi-Bit Input, Weight and Output and Achieving 351 TOPS/W and 372.4 GOPS​

Publisher: IEEE
Cite This


Mahmut E. Sinangil; Burak Erbagci; Rawan Naous; Kerem Akarvardar; Dar Sun; Win-San Khwa; Hung-Jen Liao; Yih Wang; Jonathan Chang
Sign In or Purchase
30
Paper
Citations
4250
Full
Text Views

Abstract
Authors
Figures
References
Citations
Keywords
Metrics
More Like This
Footnotes

• Download PDF
• View References
• Request Permissions
• Save to
• Alerts

Abstract:In this work, we present a compute-in-memory (CIM) macro built around a standard two-port compiler macro using foundry 8T bit-cell in 7-nm FinFET technology. The proposed design supports 1024 4 b $\times $ 4 b multiply-and-accumulate (MAC) computations simultaneously. The 4-bit input is represented by the number of read word-line (RWL) pulses, while the 4-bit weight is realized by charge sharing among binary-weighted computation caps. Each unit of computation cap is formed by the inherent cap of the sense amplifier (SA) inside the 4-bit Flash ADC, which saves area and minimizes kick-back effect. Access time is 5.5 ns with 0.8-V power supply at room temperature. The proposed design achieves energy efficiency of 351 TOPS/W and throughput of 372.4 GOPS. Implications of our design from neural network implementation and accuracy perspectives are also discussed”

I chose to Google 7nm as the former CEO Mr. Dinardo when asked by a shareholder in one of his webinars for the first time that I am aware said “Yes it can scale down from 28 to 14 to 7nm”. Since then Anil Mankar and Peter van der Made have also mentioned 4nm and 5nm respectively.

The intriguing part of Numen referencing 22nm apart from this is that Anil Mankar said in the Anastasia video that NASA was looking at 90nm. At 90nm semiconductors are more resilient to radiation and as I understand it semiconductors used in defence applications also seek similar resilience.

All I have at this stage.

My opinion only DYOR
FF

AKIDA BALLISTA

Cheers.

Is intriguing as you say though appears one thought last year was already to aim for 22nm and indicated AKD500 might be the implementation

Obviously mentioned 90nm as well which would tie in with NASA as you point out.

Akida spiking neural processor could head to FDSOI

Startup BrainChip Inc. has a road-map of larger and smaller SoC exemplars of its Akida spiking neural networking processing architecture and could go to fully-depleted silicon on insulator (FDSOI).

www.eenewseurope.com

Anil Mankar, chief development officer, told eeNews Europe: “Chip production volume is just starting now. But you will see a lot of IP licensing going forward.” He added: “We are process agnostic.”

The near-term focus is supplying the Akida IP to 22nm although some customers may go back to 90nm process, Akida executives said.

Rob Telson, vice president of worldwide sales, said BrainChip is drawing up plans for smaller and larger versions of Akida under the names Akida500, Akida1500 and Akida2000. Some of these may well comply to a new generation of the Akida architecture – Akida 2.0 – due to arrive in 2022. It is thought Akida500 could be implemented in 22nm FDSOI manufacturing process, and serve as a demonstrator of the agnostic nature of the Akida architecture.

 

[buena suerte :-)]

Macca
Concur with your comment re Dodgy knees. He is an asset to this group as are a number of great informative posters so thanks too

Yep some awesome research, time and knowledge shared generously here @tse... Also a big shout out to you @Earlyrelease & Tony Dawe for your great efforts as our 'Social events coordinators' here in WA .... many drinks and chats have been shared with fellow Chippers ... Looking forward to the next one!!

 

[alwaysgreen]

WBT up 2% today and up 50% over the past 6 months. Why? Loads of great news and the CEO recently held a "Meet and greet event" with the shareholders where he disclosed that they were confident in signing with a Tier 1 fab prior to Christmas.

BRN - hold a meeting and release some of your secrets

 

[Diogenese]

Hi Macca,

It is significant, but it is not Akida.

IMEC is analog NNs.


EP3674982A1 HARDWARE ACCELERATOR ARCHITECTURE FOR CONVOLUTIONAL NEURAL NETWORK


EP3671750A1 SYNAPSE CIRCUIT WITH MEMORY


EP3671568A1 BINARY RECURRENT NEURAL NETWORK INFERENCE TECHNIQUE


EP3671748A1 IN-MEMORY COMPUTING FOR MACHINE LEARNING


US2020210822A1 Multibit Neural Network


US2020151550A1 Machine Learning Accelerator


US2020159809A1 Convolution Engine for Neural Networks

IMEC are still in prototype stage of their Capacitive FeRAM, so I would guess it will be a few years before they are production-ready. After all, they probably have not experienced the immediate commercial urgency of manufacturers - IMEC = Interuniversity Microelectronics Centre (IMEC) .

https://www.newelectronics.co.uk/co...earer-to-developing-feram-memory-applications

“Now we have a high-performance, scalable, and CMOS-compatible ferroelectric capacitor technology that will bring us to the next exciting phase, i.e., moving from planar to 3D FeRAM capacitor structures – leveraging atomic layer deposition (ALD) processes,” added Van Houdt. “This will increase the density needed to bring FeRAM memories to the market as a new embedded or standalone memory.”

 

[rgupta]

WBT up 2% today and up 50% over the past 6 months. Why? Loads of great news and the CEO recently held a "Meet and greet event" with the shareholders where he disclosed that they were confident in signing with a Tier 1 fab prior to Christmas.

BRN - hold a meeting and release some of your secrets

Different strategies, different goals and different outcomes.
But for long term holders that does matter because results will come in the long run.
So yes if you believe then buy the dip, if donot believe then go with the market.
Brainchip have much bigger ethics than any company of its size.

 

[Diogenese]

Cheers.

Is intriguing as you say though appears one thought last year was already to aim for 22nm and indicated AKD500 might be the implementation

Obviously mentioned 90nm as well which would tie in with NASA as you point out.

Akida spiking neural processor could head to FDSOI

Startup BrainChip Inc. has a road-map of larger and smaller SoC exemplars of its Akida spiking neural networking processing architecture and could go to fully-depleted silicon on insulator (FDSOI).

www.eenewseurope.com

Anil Mankar, chief development officer, told eeNews Europe: “Chip production volume is just starting now. But you will see a lot of IP licensing going forward.” He added: “We are process agnostic.”

The near-term focus is supplying the Akida IP to 22nm although some customers may go back to 90nm process, Akida executives said.

Rob Telson, vice president of worldwide sales, said BrainChip is drawing up plans for smaller and larger versions of Akida under the names Akida500, Akida1500 and Akida2000. Some of these may well comply to a new generation of the Akida architecture – Akida 2.0 – due to arrive in 2022. It is thought Akida500 could be implemented in 22nm FDSOI manufacturing process, and serve as a demonstrator of the agnostic nature of the Akida architecture.

Now look what you made me do:

What is FDSOI anyway?

https://anysilicon.com/fdsoi/

FDSOI stands for Fully Depleted Silicon on Insulator. FDSOI is a planar process technology that provides an alternative solution to overcome some of the limitations of bulk CMOS technology at reduced silicon geometries and smaller nodes.

The FDSOI process has two distinct features. First starting with the substrate, an ultra-thin buried oxide layer is placed on the top of the base silicon. Second, a very thin silicon layer on top, creates the transistor channel. The FDSOI transistor is built above the buried-oxide layer. It has a raised source and drain and a very thin uniform channel under the gate. As the transistor channel layer is very thin, no channel doping is required, which makes the transistor fully depleted. The resulting FDSOI device is shown in the illustration in Fig. 1.1.

Fig. 1.1: FDSOI transistor

FDSOI is not only an alternative to continued CMOS scaling beyond 28nm node but also proves to be useful in improving the performances of older technology nodes when redeveloped using this process.

Two IDMs, STMicroelectronics and Renesas, also have their own FDSOI technology. STMicroelectronics adopted FD-SOI technology in 2012 and started several projects. STMicroelectronics is now using this technology for many diversified market applications. Renesas has a 65nm version of FDSOI called silicon-on-thin-BOx (SOTB), which targets ultra-low-power MCU markets. A partnership on FDSOI fabrication between ST and Samsung Foundry was announced in 2014. Today, two out of the world’s top foundries offer FDSOI technology fabrication. Samsung offers its 28nm and GF has its 22nm node fully developed for production. Many companies such as NXP, Sony, Mobileye started projects on FDSOI technology after the foundry system was in place. Today the IPs are well developed to make any sort of application possible. Both Samsung and GLOBAL FOUNDRIES are also developing its 18nm and 12nm nodes for future production, respectively. In Feb 2017, GF announced its plans to expand the capacity of its Fab 1 facility in Dresden due to market’s growing interest. In September 2017, Soitec announced to re-open its Singapore fab to meet increasing customer demand for wafers.
...
It provides improved speed and reduced power consumption* through a simple manufacturing process, so the power-performance-area-cost tradeoff is very good as compared to both bulk-CMOS and FinFet.
...
This technology is already supported by multiple foundries and IP solution providers for 28nm & 22 nm. The future roadmap 12nm is already in place and the process can be scaled down below 10 nm as well.

* Any more power reduction and Akida will become endothermal.

From "WayBack", this article appears to have been updated in November 2022.

 

[SERA2g]

In regards to our company name I would prefer Akida. It's not confusing and a lot cooler, saying that at some point you are known as something and sometimes it just sticks.

I disagree with this, respectfully of course!

Akida, what does that even mean? Prior to owning Brainchip shares, I had no idea that akida was the greek work for spike.

Now, knowing as I do know that akida is based on a spiking neural network, akida makes sense for the product name.

For the company though, as a producer artificial intelligence IP inspired by the human brain, Brainchip makes complete sense for the company name.

I don't need to know what a neural network or a spike is to understand what "Brainchip" means.

The name Brainchip has also, by pure chance, worked well for us in regards to raising awareness of the company because of it's likeness to Neuralink. You would have all seen articles about Neuralink on twitter. If you google "elon musk neuralink" there are hundreds of articles and videos which refer to Elon's chip as "brain chip".

All good reasons why Brainchip is a great company name.

 

[Fullmoonfever]

Now look what you made me do:

What is FDSOI anyway?

https://anysilicon.com/fdsoi/

FDSOI stands for Fully Depleted Silicon on Insulator. FDSOI is a planar process technology that provides an alternative solution to overcome some of the limitations of bulk CMOS technology at reduced silicon geometries and smaller nodes.

The FDSOI process has two distinct features. First starting with the substrate, an ultra-thin buried oxide layer is placed on the top of the base silicon. Second, a very thin silicon layer on top, creates the transistor channel. The FDSOI transistor is built above the buried-oxide layer. It has a raised source and drain and a very thin uniform channel under the gate. As the transistor channel layer is very thin, no channel doping is required, which makes the transistor fully depleted. The resulting FDSOI device is shown in the illustration in Fig. 1.1.


View attachment 23700


Fig. 1.1: FDSOI transistor

FDSOI is not only an alternative to continued CMOS scaling beyond 28nm node but also proves to be useful in improving the performances of older technology nodes when redeveloped using this process.

Two IDMs, STMicroelectronics and Renesas, also have their own FDSOI technology. STMicroelectronics adopted FD-SOI technology in 2012 and started several projects. STMicroelectronics is now using this technology for many diversified market applications. Renesas has a 65nm version of FDSOI called silicon-on-thin-BOx (SOTB), which targets ultra-low-power MCU markets. A partnership on FDSOI fabrication between ST and Samsung Foundry was announced in 2014. Today, two out of the world’s top foundries offer FDSOI technology fabrication. Samsung offers its 28nm and GF has its 22nm node fully developed for production. Many companies such as NXP, Sony, Mobileye started projects on FDSOI technology after the foundry system was in place. Today the IPs are well developed to make any sort of application possible. Both Samsung and GLOBAL FOUNDRIES are also developing its 18nm and 12nm nodes for future production, respectively. In Feb 2017, GF announced its plans to expand the capacity of its Fab 1 facility in Dresden due to market’s growing interest. In September 2017, Soitec announced to re-open its Singapore fab to meet increasing customer demand for wafers.
...
It provides improved speed and reduced power consumption* through a simple manufacturing process, so the power-performance-area-cost tradeoff is very good as compared to both bulk-CMOS and FinFet.
...
This technology is already supported by multiple foundries and IP solution providers for 28nm & 22 nm. The future roadmap 12nm is already in place and the process can be scaled down below 10 nm as well.

* Any more power reduction and Akida will become endothermal.

So the clues in that 2021 article were there and could be what's happening now then?

Some decent names using the process it seems....wonder who owns the third party device referenced by Chittipeddi

 

[Diogenese]

same here

Bugger again!

I'm still saving my pennies for Talga options next week.

 

[alwaysgreen]

Different strategies, different goals and different outcomes.
But for long term holders that does matter because results will come in the long run.
So yes if you believe then buy the dip, if donot believe then go with the market.
Brainchip have much bigger ethics than any company of its size.

Respectfully, I don't want dips. I am full to my limit on BRN and am now not buying any more so "dips" don't help me. Especially 16% dips in a week.

 

[Diogenese]

Respectfully, I don't want dips. I am full to my limit on BRN and am now not buying any more so "dips" don't help me. Especially 16% dips in a week.

Maybe you could tune your suspension for the corrugations until we hit the bitumen.

 

[alwaysgreen]

Maybe you could tune your suspension for the corrugations until we hit the bitumen.

 

[Straw]

I disagree with this, respectfully of course!

Akida, what does that even mean? Prior to owning Brainchip shares, I had no idea that akida was the greek work for spike.

Now, knowing as I do know that akida is based on a spiking neural network, akida makes sense for the product name.

For the company though, as a producer artificial intelligence IP inspired by the human brain, Brainchip makes complete sense for the company name.

I don't need to know what a neural network or a spike is to understand what "Brainchip" means.

The name Brainchip has also, by pure chance, worked well for us in regards to raising awareness of the company because of it's likeness to Neuralink. You would have all seen articles about Neuralink on twitter. If you google "elon musk neuralink" there are hundreds of articles and videos which refer to Elon's chip as "brain chip".

All good reasons why Brainchip is a great company name.

All fair points.

 

[buena suerte :-)]

I disagree with this, respectfully of course!

Akida, what does that even mean? Prior to owning Brainchip shares, I had no idea that akida was the greek work for spike.

Now, knowing as I do know that akida is based on a spiking neural network, akida makes sense for the product name.

For the company though, as a producer artificial intelligence IP inspired by the human brain, Brainchip makes complete sense for the company name.

I don't need to know what a neural network or a spike is to understand what "Brainchip" means.

The name Brainchip has also, by pure chance, worked well for us in regards to raising awareness of the company because of it's likeness to Neuralink. You would have all seen articles about Neuralink on twitter. If you google "elon musk neuralink" there are hundreds of articles and videos which refer to Elon's chip as "brain chip".

All good reasons why Brainchip is a great company name.

Totally agree mate it's a great name ... and I'm slightly bias of course.. BRN 1

 

[DollazAndSense]

Well the newest podcast with Sally was certainly interesting. To me, it seems like she is quite hesitant to acknowledge the technological achievement of Akida as little more than a ‘niche’ product.

Not phased one little bit about a journalists opinion but Sally came across as quite dismissive. This is to be expected however and shows the challenge Brainchip’s team faces in selling our technology to customers. We’ll get there but it seems patience is going to be required for quite a while yet.

 

[Fact Finder]

In regards to our company name I would prefer Akida. It's not confusing and a lot cooler, saying that at some point you are known as something and sometimes it just sticks.

Though a quick Google could cause someone to think it is a canine or a Japanese cartoon character.

The reality is that AKIDA is being sold to industry who will not be confused by the name Brainchip Inc and will in most cases for many years not see commercial value in proudly proclaiming that the brilliance of their product is down to some other companies Science Fiction.

My opinion only DYOR
FF

AKIDA BALLISTA

 

[HopalongPetrovski]

I disagree with this, respectfully of course!

Akida, what does that even mean? Prior to owning Brainchip shares, I had no idea that akida was the greek work for spike.

Now, knowing as I do know that akida is based on a spiking neural network, akida makes sense for the product name.

For the company though, as a producer artificial intelligence IP inspired by the human brain, Brainchip makes complete sense for the company name.

I don't need to know what a neural network or a spike is to understand what "Brainchip" means.

The name Brainchip has also, by pure chance, worked well for us in regards to raising awareness of the company because of it's likeness to Neuralink. You would have all seen articles about Neuralink on twitter. If you google "elon musk neuralink" there are hundreds of articles and videos which refer to Elon's chip as "brain chip".

All good reasons why Brainchip is a great company name.

I too like Brainchip although would prefer it BrainChip.
The main thing I like about it is, it's just so easy to remember.
It is a concept which already has an innate depth of meaning (at least in the west) and is also a little science fiction which adds to the memorability and also lends it a certain panache and geek sheik.
It is also a universally understandable conception in one word and was indeed the first thing to grab my curiosity when I first stumbled upon it, a 'coons age ago.
You don't need to be a nerd or highly technically literate to appreciate an easy idea of just what our Company is going for.
Granted, a little research will soon perhaps dispel the idea that we are not in actuality Cyberdyne incarnate, but by then many are hooked by the perhaps dimly grasped potential and the vague understanding that one may just have stumbled upon a next generation behemoth in the making.

I also like AKIDA as a name for our first line of product, although personally wish they could have kept our synapse logo.

I understand the more "perhaps" corporate rebrand and whilst I appreciate its "professionalism" liked the whimsy of the logo which is not language specific much like Apple's now eponymous symbol.

I think Elon has a touch of the marketing genius in the names he chooses to represent his organisations....
PayPal, Tesla, Space X, Neuralink etc.... again, all somewhat recognisable at first glance and indeed first contact.

I am glad that with Brainchip, we have stolen a march on him, in this instance.

 

[Diogenese]

Though a quick Google could cause someone to think it is a canine or a Japanese cartoon character.

The reality is that AKIDA is being sold to industry who will not be confused by the name Brainchip Inc and will in most cases for many years not see commercial value in proudly proclaiming that the brilliance of their product is down to some other companies Science Fiction.

My opinion only DYOR
FF

AKIDA BALLISTA

... but now that Mercedes has shattered the zone of silence, I think that once other customers are near production, they will want to take advantage of Akida as a selling point.

"Coming soon, Tandoorator, the AI refrigerator that mixes your biryani while you munch chappati- AKIDA INSIDE!"

 

[David holland]

Well the newest podcast with Sally was certainly interesting. To me, it seems like she is quite hesitant to acknowledge the technological achievement of Akida as little more than a ‘niche’ product.

Not phased one little bit about a journalists opinion but Sally came across as quite dismissive. This is to be expected however and shows the challenge Brainchip’s team faces in selling our technology to customers. We’ll get there but it seems patience is going to be required for quite a while yet.

Couldn't agree more certainly wasn't on there to piss in the pockets of brainchip holders

 

[HopalongPetrovski]

... but now that Mercedes has shattered the zone of silence, I think that once other customers are near production, they will want to take advantage of Akida as a selling point.

"Coming soon, Tandoorator, the AI refrigerator that mixes your biryani while you munch chappati- AKIDA INSIDE!"