BRN Discussion Ongoing

[Fullmoonfever]

This came up as featured in the latest edition and link takes you to the below article which may have been posted before?

BrainChip tapes out its AI chip in 22nm FD SOI

BrianChip has taped out its AKD1500 event based machine learning chip on the low power GlobalFoundries 22nm FD-SOI technology.

www.eenewseurope.com

BrainChip has taped out its latest AKD1500 event based machine learning chip on the low power GlobalFoundries 22nm fully depleted silicon-on-insulator (FD-SOI) technology.

The reference design is a key milestone is part of validating BrainChip’s IP across different processes and foundries. It has also been working with Renesas on integrating the IP, called Akida, into a chip that taped out last month.

• Renesas tapes out spiking neural network chip
• NASA looks to BrainChip’s spiking neural network chip
• Akida spiking neural processor could head to FDSOI

The AKD1500 uses the latest features of BrainChip’s AI IP in combination with the ultra-low leakage FDX22 process for always-on, at-sensor applications or other low-power embedded applications in AIoT, industrial, consumer and automotive markets.

For always-on AI at the sensor, Akida can offload feed-forward networks without the use of a CPU, which enables extremely energy efficient intelligent sensors. For other low-power embedded applications, the CPU solely performs sensor fusion processes or runs the application with Akida managing all the AI acceleration to significantly minimize CPU workload and system load, freeing up the MCU for other general-purpose computation.

Spiking neural networks​

The AKD1500 is designed as an accelerator reference chip, using quad/octal SPI interfaces for MCU modules and PCIe for extensible cards, which assists partners in developing and demonstrating their solutions as a stepping stone to integrating the Akida IP into their production SoCs.

• BrainChip joins Intel Foundry Services to drive neuromorphic AI
• Brainchip ships first mini PCIexpress board with spiking neural network chip
• BrainChip provides details of neural network architecture

“This is an important validation milestone in the continuing innovation of our event-based, neuromorphic Akida IP, even if it is fully digital and portable across foundries,” said Anil Mankar, Co-Founder & Chief Development Officer at BrainChip. “The AKD1500 reference chip using GlobalFoundries’ very low-leakage FD SOI platform, showcases the possibilities for intelligent sensors in edge AI.”

 

[manny100]

An investment in BRN is an investment in the future. Interestingly according to the 'fireside chat' AKIDA makes current generic wearables obsolete. AKIDA will learn to gather information about what is normal and abnormal for yourself and provide feedback.
Another example was the recent is rhe recent AI labs US ann. $USS50 billion is lost in unplanned machine stoppages per year. AKIDA predictive maintenance could potentially save a lot of money.
The increasing volume and complex 'computes going through the cloud via Data Centers is unsustainable. AI at the Edge will solve this problem.
Still no sales revenue on the horizon yet. - largely due to the World just starting to be aware of AI at the Edge.
ChatGPT is baby' toys' compared to AKIDA.
In any case i am patient and prepared to wait. I am still accumulating.

 

[Learning]

Learning 🏖

 

[Slymeat]

I was hoping to hear something concrete on LSTM/transformers.

Same here.

 

[manny100]

I do not mind the shorters at high levels. It gives me a chance to accumulate a few more.
I have been in so long that i have become immune to the Share price volatility.

 

[DingoBorat]

12 car companies caused autonomous driving accidents last year and we only hear about Tesla.
Ideally, yes, we would be better of removing human error element from the roads, but we are way off from it.

So, at this stage, it is drivers responsibility to take care on the road, even if he/she uses autonomous feature, be ready to respond to emergency.

I'm still against these journos vendetta against one person and company who leads the world in electric vehicles and what they offer.
No different to some people like Nanalyze who talks always rubbish about Brainchip because it hasn't been taken over yet.

Journos are journos, it's not a vendetta, they're just doing the stories that will provide the most clicks (to earn advertising dollars)..

This side of journalism, in all areas, is sickening.

 

[Fullmoonfever]

I do not mind the shorters at high levels. It gives me a chance to accumulate a few more.
I have been in so long that i have become immune to the Share price volatility.

This one's for you then

 

[TheFunkMachine]

I find the complete absence of any information about LSTM & Transformers in this podcast extremely exciting.

Previously Peter van der Made said as Acting CEO AKD 2000 was ready to go to engineering shortly.

Ken Scarince explaining the payments recorded in the half yearly said they had bought and paid for all the necessary third party IP for AKD 2000.

Rob Telson was excited in the Carnegie Mellon podcast about transformers for natural language processing.

Anil Mankar said at 2021 Ai Field day there would be a little bit of LSTM in AKD1500.

As a matter of proven science and engineering there are multiple papers supporting that CNN, RNN & SNN are compatible with LSTM and Transformers.

NOW COMPLETE SILENCE. THIS SILENCE HAS TO BE SIGNIFICANT in my opinion.

As for whether this Podcast was just fluff I will remind everyone that various posters here have lauded the ability of Peter van der Made to set out a timetable for the completion of his vision and marvelled that since he did so back in 2015 he has met every technology development target yet those same posters have not given Peter van der Made's statement that AGI of the type he described will be achieved by Brainchip in 2030 approximately. If Bill Gates made the same statement imagine how the World's press would respond.

The third and compelling piece of information in this Podcast comes from the clear statement that by producing AKD1500 without a CPU in fully digital format in 22nm at Global Foundries Brainchip will be swiping the feet out from under all those claiming to be able to produce SNN in analogue as no one developing analogue SNN for the edge can offer such flexibility and scalability because of the yet to be overcome difficulties involved in producing analogue chips.

@Diogenese has expanded on this numerous times and multiple papers have been posted together with links on HC and TSEx. This revealed FACT is SIGNIFICANT and those suggesting it is a fluff podcast clearly have no idea about AKIDA technology revolution and should refrain from commenting until they do. SCIENCE is not about lay opinions.

While not spoken of in great detail at 22nm from Global Foundries the reduction of leakage involved in their design ordinarily produces up to a 70% improvement in power use of a traditional von Neumann designed chip.

This is a well known and understood FACT that anyone and everyone in the semiconductor space just knows so when all those on the Podcast glossed over this FACT and spoke to the huge power saving advantage of AKIDA technology they most likely thought that even your average retail investor in Brainchip would understand what a further up to 70% power saving is not fluff and will mean that when Brainchip presents the AKD1500 to the World it will create ripples throughout the semiconductor industry.

The already inconceivable science fiction of AKD1000 at 300 megahertz outperforming a GPU running at 900 megahertz just gets blown out into something that is beyond Edge Impulses understanding of what constitutes science fiction. Imagine the reaction that Nviso will have to what AKD1500 can do????

This is my opinion only so feel free to urinate all over these known provable FACTS for your own malicious purposes and continue to claim this was just a fluff Podcast.
FF

AKIDA BALLISTA

Welcome back FF!

My personal favourite in this podcast is when they talk about the edge is just the beginning for brainchip and that they will be catching a lot of the cloud computations done today. Did I hear that right?

AGI by 2030 is insane, Akida 10 is the name. We have Akida 1 and Akida 1,5 currently, which in my mind means that by 2030 we will have at least 8 new releases in the next 7 years with a steady incline in capabilities and performance. Watch out.

It is also stated that the purpose for akd1000 is mainly to prove its viability and for as many companies/individuals as possible the opportunity to easily prototype new applications.

In one point of the podcast it was said by Peter that Akd1500 has a much smaller footprint and much more efficient or something to that extent. This blows my mind. We all out here thinking akd1000 is si-fi, whilst they only claim it to be a good entry point for people to play around and get excited.

What’s to come from brainchip is where agreements and costumers will come flooding imo.

I have just decided that I won’t touch my shares for 3 years to see how right they are.

 

[jtardif999]

To me it was nothing more than just a feel good "Fluff " podcast of things to come well well down the possible future track .... might of as well given us all an idea of what will come in say 20 years time with Akida20.
I also just like some others here would have much preferred to have heard more factual information about the current state of affairs with the LTSM ... Akida 1500 situation without breaching any NDA's.

There may be a reason why they did not mention LSTM and Transformers ats. Perhaps the release of the Akida2.0 IP is just around the corner and good gosh they plan to make an ASX announcement about that. I think as has been discussed by others here, AKD1500 probably doesn’t have LSTM and Transformers. My gut feel is that that will be released with the updated architecture and that this time it’s about suiting particular client(s) for particular products with extreme low power budgets. Anyway all will be revealed in the fullness of this year I think.

 

[Slade]

The issue here for me is two fold:

1. I am invested in Brainchip. Brainchip is partnered with Mercedes Benz. Mercedes Benz has an established highly advanced system for recording accidents involving Mercedes vehicles still under warranty.

2. Tesla is not partnered with Brainchip and is a competitor of Mercedes Benz.

As a result it is not hatred of Tesla but a belief in traditional automotive practice as represented by Mercedes Benz and most importantly a successful partnership between Brainchip and Mercedes Benz will make me money. Defending the competition (Tesla) when Mercedes Benz does not even get a mention in passing just seems counter intuitive from the perspective of being an investor in Brainchip unless you are not:

“Automated tech factored in 392 car crashes in 11 months ...

15 June 2022 — Automakers reported nearly 400 crashes of vehicles with partially automated driver-assist systems, including 273 involving Teslas..”

So I do not hate Tesla. I dislike Elon Musk. Claiming that Tesla is being targeted for political reasons does not stand up against the published facts.

If Tesla partners with Brainchip and makes me money I will speak in favour of Tesla but until then they are a competitor with an inferior product.

I will always dislike Elon Musk.

My opinion only DYOR
FF

AKIDA BALLISTA

I got it!
Now I know who took a dump in front of the fire place.

 

[Fullmoonfever]

Something I've found interesting with regards to the BMW posts.

BMW exploring with SynSense Speck currently as we know.

SynSense have a partnership with Prophesee as we know.

Couple of excerpts from EE-Times EU Sept 22 and Prophesee own Ann Oct 21.

Curious as to SynSense / BMW actually using Speck with IniVation cameras instead of Prophesee considering what SynSense and Prophesee agreed to....can only assume IniVation a BMW directive maybe as I (wrongly) assumed Speck was part of the Prophesee dev.

https://www.eetimes.eu/wp-content/uploads/2022/09/EE-Times-Europe-Magazine-September-2022.pdf

Building an artificial system based on spiking isn’t easy. Not only do we not know exactly how the neuron works, but there is no agreement on the best way to train spiking networks. Backpropagation — the algorithm that makes training deep-learning algorithms possible today — requires computation of derivatives, which is not possible for spikes.

Some working in the field, such as SynSense, approximate derivatives of spikes in order to use backpropagation. Others use a technique called spike-timing–dependent plasticity (STDP), which is closer to how biological brains function; BrainChip uses this method for oneshot learning at the edge. STDP, however, is a less mature technology. There is also the possibility of taking deep-learning convolutional neural networks (CNNs) that have been trained by backpropagation in the normal way and converting them to run in the spiking domain. BrainChip is pursuing this technique as well.

SYNSENSE SPECK SynSense and BMW are collaborating to advance the integration of neuromorphic chips into smart cockpits and explore related fields. BMW will be evaluating SynSense’s Speck system-on-chip, which combines SynSense’s neuromorphic vision processor with a 128 × 128-pixel event-based camera from iniVation. The SoC can be used to capture real-time visual information, recognize and detect objects, and perform other vision-based detection and interaction functions.

Prophesee

SynSense and Prophesee develop one-chip event-based smart sensing solution

Partnership leverages respective leadership in neuromorphic computing and sensing to realize on-sensor processing integration leading to small form-factor, low-power and cost-effective IoT solutions

www.prophesee.ai

The partnership will address the design, development, manufacturing and commercialization of the combined neuromorphic technology, including sensors, processing solutions, software and solutions to enable a broad range of applications

 

[Fullmoonfever]

Found this report / book by River Publishers from July 22.

Whilst quite broad based and technical, and couldn't see direct references to BRN / Akida, it appears to have a quite a bit on neuromorphic and in depth look at the AI industry and tech overall.

Some might enjoy the skim through for other insights and others like @Diogenese for the tech info?

 

[BaconLover]

I need some help. I want to buy more BrainChip shares but I want them for under 60 cents. I’m trying to write a message that will cast doubt over any progress that BrainChip has been making. I will post it just before the market opens. It needs to distract people from the $1.2 million that BrainChip received from customers over the last three months, to take their minds away from the extra sales and marketing executives that they have hired including the one for Korea, and to hide the new breakthrough partnership with Intel. I want the tapping out the new Akida 1.5 to be a distant memory along with the Akida inspired Renesas chip that will come out later this year. BrainChip have made so much progress that it’s going to be hard to cover it up. I will target the CEO, that always scares people. I don’t want members to think I am downramping so I will write it in a passive way. I tried yesterday by attacking the decision making of the board with regards to what they choose to send to the ASX for announcing . Will you all help me. I have $130 that I want to invest. I need a really cheap price. I’m worried that the share price bottomed out yesterday and I missed my opportunity.

Hope you're loading up with that $130!

 

[BaconLover]

I like Elon, he's a visionary, has the drive to do some amazing things, and Tesla have achieved much more than Mercedes over the last few years.

Pretty sure Elon is heartbroken to hear that he is disliked by a few important peeps.

 

[wilzy123]

I got it!
Now I know who took a dump in front of the fire place.

The cooked have been put on notice

 

[Food4 thought]

I find the complete absence of any information about LSTM & Transformers in this podcast extremely exciting.

Previously Peter van der Made said as Acting CEO AKD 2000 was ready to go to engineering shortly.

Ken Scarince explaining the payments recorded in the half yearly said they had bought and paid for all the necessary third party IP for AKD 2000.

Rob Telson was excited in the Carnegie Mellon podcast about transformers for natural language processing.

Anil Mankar said at 2021 Ai Field day there would be a little bit of LSTM in AKD1500.

As a matter of proven science and engineering there are multiple papers supporting that CNN, RNN & SNN are compatible with LSTM and Transformers.

NOW COMPLETE SILENCE. THIS SILENCE HAS TO BE SIGNIFICANT in my opinion.

As for whether this Podcast was just fluff I will remind everyone that various posters here have lauded the ability of Peter van der Made to set out a timetable for the completion of his vision and marvelled that since he did so back in 2015 he has met every technology development target yet those same posters have not given Peter van der Made's statement that AGI of the type he described will be achieved by Brainchip in 2030 approximately. If Bill Gates made the same statement imagine how the World's press would respond.

The third and compelling piece of information in this Podcast comes from the clear statement that by producing AKD1500 without a CPU in fully digital format in 22nm at Global Foundries Brainchip will be swiping the feet out from under all those claiming to be able to produce SNN in analogue as no one developing analogue SNN for the edge can offer such flexibility and scalability because of the yet to be overcome difficulties involved in producing analogue chips.

@Diogenese has expanded on this numerous times and multiple papers have been posted together with links on HC and TSEx. This revealed FACT is SIGNIFICANT and those suggesting it is a fluff podcast clearly have no idea about AKIDA technology revolution and should refrain from commenting until they do. SCIENCE is not about lay opinions.

While not spoken of in great detail at 22nm from Global Foundries the reduction of leakage involved in their design ordinarily produces up to a 70% improvement in power use of a traditional von Neumann designed chip.

This is a well known and understood FACT that anyone and everyone in the semiconductor space just knows so when all those on the Podcast glossed over this FACT and spoke to the huge power saving advantage of AKIDA technology they most likely thought that even your average retail investor in Brainchip would understand what a further up to 70% power saving is not fluff and will mean that when Brainchip presents the AKD1500 to the World it will create ripples throughout the semiconductor industry.

The already inconceivable science fiction of AKD1000 at 300 megahertz outperforming a GPU running at 900 megahertz just gets blown out into something that is beyond Edge Impulses understanding of what constitutes science fiction. Imagine the reaction that Nviso will have to what AKD1500 can do????

This is my opinion only so feel free to urinate all over these known provable FACTS for your own malicious purposes and continue to claim this was just a fluff Podcast.
FF

AKIDA BALLISTA

You Fact Finder are the goods

No only are you spot on with your words but concise in your opinion.

I personally loved the podcast

Thank goodness your back

 

[Easytiger]

I'm sorry but l have decided to express my opinion about you and others and your claim that everyone has a right to their opinion think its a hissy fit.

So if it was as you say not malicious what is your excuse?

Ignorance of the FACTS and what Brainchip has achieved and a willingness to say anything that pops into your head because well what does it matter if you express opinions of no substance?

Go and scold someone who cares two hoots about your right to free speech.

I am invested in this company for one reason which is to make a profit. Your opinions have no value unless supported by FACTS.

As such I will not tolerate the argument from you and others that 'I have no idea what this is all about, I have done little to no research, I did not understand what I was reading anyway' but I am going to say this is fluff because it is my constitutional right to speak my mind no matter how it affects your investment because this is my opinion.

I have set out why this Podcast is light years away from being fluff. I invited you to urinate on my argument. Saying I have had a hissy fit and claiming you have a right to an opinion is not a response to the stated FACTS.

Either put your contrary FACTS or skulk back under your rock.

My opinion only FF

AKIDA BALLISTA

FF is back, and I’m buying tomorrow!

 

[Tothemoon24]

Skip to content
Toggle Navigation Toggle Search
Home / News & Events / GF Press Releases / GlobalFoundries Acquires Renesas’ Non-Volatile Resistive RAM Technology to Proliferate IoT and 5G Applications

GlobalFoundries Acquires Renesas’ Non-Volatile Resistive RAM Technology to Proliferate IoT and 5G Applications​

February 9, 2023
Technology acquisition expands GF’s portfolio and differentiated roadmap while accelerating the commercialization of memory solutions
MALTA, N.Y., February 9, 2023 – GlobalFoundries (Nasdaq: GFS) (GF) today announced that it has acquired Renesas Electronics Corporation (Renesas)’s proprietary and production proven Conductive Bridging Random Access Memory (CBRAM) technology, a low power memory solution designed to enable a range of applications in home and industrial IoT and smart mobile devices.
The transaction further strengthens GF’s memory portfolio and extends its roadmap of embedded non-volatile memory (NVM) solutions by adding another reliable, customizable embedded memory solution that is relatively easy to integrate into other technology nodes. Specifically, this technology will enable customers to further differentiate their SoC designs and advance a new generation of secure and intelligent devices.
“We’re committed to differentiating our technology portfolio to be the foundation of our customers’ energy efficient IoT applications today and for decades to come,” said Mike Hogan, chief business officer of GF. “With the acquisition of this innovative memory technology, GF is now playing an essential role in accelerating development of NVM solutions which will enable our customers to design the next generation of intelligent and connected devices. CBRAM technology unleashes a new paradigm of performance and ultra-low energy use, enabling a wide range of applications, from wearable devices to smartphones, to extend the time between battery charges from hours to weeks to years in specific use cases.”
CBRAM’s low power consumption, high read/write speeds, reduced manufacturing costs and tolerance for harsh environments make it particularly suitable for consumer, medical, and select industrial applications. In 2020, GF entered into a licensing agreement with Dialog Semiconductor, which was acquired by Renesas in 2021, to offer its CBRAM technology as an embedded, NVM option. Today, CBRAM is being qualified on the company’s 22FDX® platform, with plans to extend it to other platforms.

About GF​

GlobalFoundries (GF) is one of the world’s leading semiconductor manufacturers. GF is redefining innovation and semiconductor manufacturing by developing and delivering feature-rich process technology solutions that provide leadership performance in pervasive high growth markets. GF offers a unique mix of design, development and fabrication services. With a talented and diverse workforce and
an at-scale manufacturing footprint spanning the U.S., Europe and Asia, GF is a trusted technology source to its worldwide customers. For more information, visit www.gf.com.
©GlobalFoundries Inc., GF, GlobalFoundries, the GF logos and other GF marks are trademarks of GlobalFoundries Inc. Or its subsidiaries. All other trademarks are the property of their respective owners.

Forward-looking Information​

This news release may contain forward-looking statements, which involve risks and uncertainties. Readers are cautioned not to place undue reliance on any of these forward-looking statements. These forward-looking statements speak only as of the date hereof. GF undertakes no obligation to update any of these forward-looking statements to reflect events or circumstances after the date of this news release or to reflect actual outcomes, unless require by law.

Media Contact:​

Erica McGill
erica.mcgill@gf.com
+1-518-795-5240

Limitless insight with a global point of view​

Foundry Flash, our monthly email newsletter, leverages the advantage of our worldwide perspective in considering the latest industry news, solutions, technology and events
Subscribe now

• Resource Center
• Investor Relations
• Privacy Policy
• CCPA
• Terms and Conditions
• Trademarks
• Intellectual Property
• Contact Us
• Portals
• 简体中文
• German

© Copyright 2023

• Instagram
• Linkedin
• Twitter
• Facebook

 

[Tothemoon24]

Just a matter of time until I’m eating steak again

Electronics and Semiconductors

Electronica 2022: Renesas enters the automotive radar market​

Peter Brown
16 November 2022

Renesas plans to combine its radar transceiver with other semiconductors for radar system development. Source: Renesas
At Electronica 2022, taking place this week in Munich, Germany, Japanese semiconductor maker Renesas Electronics Corp. has entered the automotive radar market with the introduction of a 4x4-channel, 76-81 GHz radar transceiver.


The semiconductor is designed to meet advanced driver assistance systems (ADAS) and Level 3 and higher autonomous driving applications.
The radar transceiver, called RAA 270205, will be incorporated into its sensor fusion portfolio that combines radar, vision systems and other sensing modalities.
The semiconductor was designed with Steradian Semiconductors Private Limited, a company that Renesas acquired earlier this year.
The transceiver monolithic microwave integrated circuit (MMIC) is suited for imaging radar, long-range forward-looking radar, 4D radar and could be used for satellite automated radar systems.
Features of the RAA270205 include:

• Accuracy with up to 5 GHz of bandwidth
• A 112.5MSPS analog-to-digital converter (ADC)
• Power consumption of 1.2W
• Noise figure of 9 dB

Renesas said it plans to combine the RAA270205 transceiver with other devices from its portfolio that support automotive radar. These systems will include a satellite radar system for automated driving and ADAS in the first half of 2023.
To contact the author of this article, email PBrown@globalspec.com

 

[JK200SX]