BRN Discussion Ongoing

[rgupta]

BRN is more than 12 years old, but How old is neuromorphic technology and the first working product?

Anyway we all see things in different ways and only time will tell who’s right and who’s wrong

Donot take me wrong. I want you to be right as it will benefit as well. But your or my wish does not matter if the results are not delivered by the management.
I still assume it would had been better to bring akida 2000 than akida 1500 even if that means extra time.
I am worried company is launching akida 1500 just like akida 1000 in a hurry and any failure here will be detrimental.
On the positive note if company can recover the cost of akida 1500 from its sale it will be a good outcome.

 

[Papacass]

Harsh reality is that we could not sell enough chips or ips with akida 1000
But yes we got a line in our profile 1st neurophonic chip producer. We come in front of Intel and IBM as they cannot launch their product. The strategy only helped the day traders and shorters while the long term holders are crying.

I think you are jumping the gun re not selling enough AKD1000 IP (we don’t sell chips). Renesas is producing a product with AKD 1000 IP. Megachips also may produce something as they have a licence. Socionext and Valeo are also in the mix. We have all been educated on the design cycle. 3 to 4 years yeah? We are pretty much in the AKD 1000 sweet spot now as far as potential products go. Dry your tears and let’s see what happens.

 

[robsmark]

Donot take me wrong. I want you to be right as it will benefit as well. But your or my wish does not matter if the results are not delivered by the management.
I still assume it would had been better to bring akida 2000 than akida 1500 even if that means extra time.
I am worried company is launching akida 1500 just like akida 1000 in a hurry and any failure here will be detrimental.
On the positive note if company can recover the cost of akida 1500 from its sale it will be a good outcome.

I think you have this a little twisted rgupta.

Akd1000 - was released to customers (primarily those in the EAP) for incorporation into their intended products.

It was also licensed by two non-EAP customers.

These EAP customers requested some changes, which have been incorporated into Akd1500, a chip that’s basically a tweaked version of the original, with the same intended uses.

Akd2000 is a different beast all together and by large would be an overkill for those originally use cases.

Similar to how all the other chip manufacturers (including ARM) have many different offerings for different applications.

The existence of one doesn’t relinquish the value of another. You wouldn’t need the same processor for AI doorbell as you would for a LiDAR system for example.

 

[toasty]

I'm still a bit puzzled by Akida1500, mainly the reason for its existence. Did ARM ask for it so that they could continue to sell their current CPU's but add Akida on? Did Mercedes ask for it as they have decided to use one or more other CPU's and didn't see the need for another one onboard with Akida? But if we're selling IP instead of silicon how does that work? Our IP needs to go into some form of silicon for it to work doesn't it?? I may be slow but its all a little hazy for me.

An explanation from the company (in laymans terms) would, I think, calm the market. In the absence of information people tend to fill the void with fears and uncertainty.

 

[Labsy]

I just put in an order. trying to pinch a few at 28 cents. Hopefully it will pay off in the short to medium term.

Same

 

[Papacass]

Also, one more thing. I’ve been a BRN share holder for 6 or 7 years now so I’m a little punchdrunk. It’s hard keeping the gloves up constantly haha. Anyway, from memory, didn’t Megachips supply some input into the AKD1500 design? So 1500 is out now. PVDM said we all underestimated the importance of the Megachips deal. Maybe PVDM’s meaning will crystallise now. I’m putting my mouth guard back in for another day on the BRN express. Onward.

 

[jk6199]

Also, one more thing. I’ve been a BRN share holder for 6 or 7 years now so I’m a little punchdrunk. It’s hard keeping the gloves up constantly haha. Anyway, from memory, didn’t Megachips supply some input into the AKD1500 design? So 1500 is out now. PVDM said we all underestimated the importance of the Megachips deal. Maybe PVDM’s meaning will crystallise now. I’m putting my mouth guard back in for another day on the BRN express. Onward.

I think I'll just stay drunk until the price rises

 

[Shadow59]

I think I'll just stay drunk until the price rises

You know you'll end up with liver damage

 

[jk6199]

You know you'll end up with liver damage

Banking on BRN for a replacement one if this keeps up

 

[Cartagena]

I wish I had more bloody funds. This share price is beyond a joke.

It definitely is. I've also exhausted my funds and deeply red The opportunity cost of being in red also frustrates me. I could have made an easy $10k on 4DS but decided to put it into Brainchip at 32cents.
4DS is not even making revenue yet it's booming. Something seriously wrong here!!!

 

[mrgds]

Hi Tech,

It's interesting that the 1500 has 8 nodes (32 NPUs). This will be capable of supporting fairly powerful NNs, and, in addition, for tasks requiring more than 8 nodes, Akida's ability to dynamically reprogram for a second or further run through the nodes can be demonstrated.

What has TESLA Discovered ?????????????????

Gday @Diogenese
Your reply to @TECH regarding Akd 1500 having 8 nodes and being capable of supporting powerful NNs has me wanting to ask of your opinion here.

Recently Tesla announced they have undertaken a "different method of achieving FSD ". It was disclosed by Tesla that they had reduced the "coding used " by 99% -- ie from 300,000 lines down to 3,000.

Now, within the last 48hrs Elon Musk did a livestream on X ( posted below) of a Tesla Model S driving on FSD without any code program, relying soley on "camera vision & Ai ".

Comments made during the video by Musk,
# Using 8 nodes/Python
# No pre-mapping/routing, just NNs
# No Internet connection required
# Through vision alone has learnt how to drive, just like a human would.

My thoughts/questions are, has this ability been the product of SMART SENSORS/CAMERAS ?
ONE SHOT LEARNING ?
INFERENCE AT THE EDGE ?
LOW POWER CONSUMPTION ?

And of course, "bring out the Ogre " if none of this IYO makes you think that there isn"t any chance of Akida technology being at the heart of this.

MIND BLOWING 10MIN VIDEO....................... DEFINATELY WORTH A WATCH

REMEMBER "HEY HEY " SAYING .................."ONCE YOUR IN, .....YOUR IN"

AK"IN"DA BALL"IN"STA

 

[rgupta]

I think you are jumping the gun re not selling enough AKD1000 IP (we don’t sell chips). Renesas is producing a product with AKD 1000 IP. Megachips also may produce something as they have a licence. Socionext and Valeo are also in the mix. We have all been educated on the design cycle. 3 to 4 years yeah? We are pretty much in the AKD 1000 sweet spot now as far as potential products go. Dry your tears and let’s see what happens.

But the same is not what Antonio was telling us. There was not even a single positivity about akida1000 but rather he told us the partners put us on hold until akida 2000 is coming.

 

[Cartagena]

I'm still a bit puzzled by Akida1500, mainly the reason for its existence. Did ARM ask for it so that they could continue to sell their current CPU's but add Akida on? Did Mercedes ask for it as they have decided to use one or more other CPU's and didn't see the need for another one onboard with Akida? But if we're selling IP instead of silicon how does that work? Our IP needs to go into some form of silicon for it to work doesn't it?? I may be slow but its all a little hazy for me.

An explanation from the company (in laymans terms) would, I think, calm the market. In the absence of information people tend to fill the void with fears and uncertainty.

I strongly believe that when Brainchip sell an IP license or contract then there must be a clear clause in the contract that it must be used in an end product. That's a must. We can't afford for customers to shop around anymore, have access to our valuable IP for their own goals and tyre kick. Deals must be done. That's business.

 

[Diogenese]

Cheif scientist at BRN likes it and not for the first time....

Something tells me perhaps the new "analog" ai chip from ibm may not be that analog......

View attachment 43171

This is an IBM patent application which claims to adapt analog ReRAM for AI applications. It uses a hybrid analog/digital technique.

US2023206964A1 DIGITAL PHASE CHANGE MEMORY (PCM) ARRAY FOR ANALOG COMPUTING 20211229

A plurality of bit lines corresponding to elements of an input vector intersect a plurality of word lines and a plurality of memristive cells are located at the intersections. At least three cells are grouped together to represent a single matrix element. At least three word lines correspond to each element of an output vector. An A/D converter is coupled to each of the word lines, and for each line, except a first, in each group, a shifter has an input coupled to one of the A/D converters. For each group, an addition-subtraction block adds the output of the A/D converter coupled to the first one of the word lines to outputs of each of the shifters except that for a last one of the word lines, subtracts the output of the last shifter, and outputs a corresponding element of an output vector.

[0045] One or more embodiments thus provide a non-volatile memory circuit 100 , including an array of storage units 108 connected along a plurality of bit lines 101 - 1 , 101 - 2 , . . . , 101 -m - 1 , 101 -m. Each storage unit 108 is configured to store one bit of information. To form a single weight (i.e., a single element of the weight matrix such as G11 103 or G12 105 ), n storage units 108 are taken together. The kth storage unit of the n storage units represents the 2k−1 place. So, for G11 labeled 103 , g1,1,0 corresponds to 21-1 or 20 or ones (least significant bit), g1,1,1 , corresponds to 22-1 or 21 or twos, . . . , and g1,1,n−1 corresponds to 2n−1 (most significant bit, k=n). A multi-bit ADC converter 109 is connected to each column output to convert the current output to a digital signal. A shifter 111 - 1 , . . . , 111 -n− 2 , 111 -n− 1 is connected to each ADC output (except the first column for a given entry in the weight matrix) to shift the digital signal by 2k−1 . The bit shifters for each column output are used in one or more embodiments for further digital operation. Digital adders and a digital subtractor are respectively used for every n columns to sum the input from columns 1 to n−1 (or 0 to n−2) and to subtract the nth column (or n−1th column). Each group of n columns 0 . . . n−1 (or 1 . . . n) has multiple adders 113 and a subtractor 115 . The digital adders sum the input from columns 1 to n−1 (or 0 to n−2) and the subtractor 115 subtracts the nth (or n−1th ) column; the output of subtractor 115 is the corresponding element in the Y vector, i.e., Y1 , Y2 , . . . , Yi . The skilled artisan can choose what indices to employ; e.g., 1 to n or 0 to n−1. A single sign bit is used in one or more embodiments to map the weight to the negative region.
...
[0049] Furthermore, one or more embodiments use a binary resistive memory array for analog computing, which is significantly faster and more energy efficient than digital computing.

[0050] Thus, one or more embodiments advantageously employ an additional bit shifter to quickly add up the weights from different columns.

[0051] One or more embodiments employ current-domain computation relying on conductance of the individual memory cells 108 . One or more embodiments use resistive memory which is non-volatile, occupies less footprint than static random access memory (SRAM), and consumes less energy. One or more embodiments employ the positional number system. For example, a number of prior art techniques do not use the positional number system for digital weight; instead, they employ the memristor as a continuous analog weight. In contrast, one or more embodiments employ digital weights with the positional number system, which greatly mitigates the non-ideality that is inherent with analog weights, such as conductance drift, nonlinear conductance update, retention, noise, and the like.

[0052] Once the weights are stored in the array, inference can be performed. Non-limiting examples of inference include handwriting recognition, such as recognition of hand-written digits; classification of images; and natural language processing. Furthermore with regard to storing weights, see discussion elsewhere herein re pulse units for programming the cells with weights by causing changes in states of the phase change material.

PS: My eyesight isn't quite good enuf to see if it's the same chip:

 

[miaeffect]

I agree 100% tech takes time to establish but here the question is how you plan and implement things.
On the time frames brn is more than 12 years old as well and still no profits.

The other point I want to raise here is every good product cannot be a success just because it is a good product. We you give a good product to bad manager they can kill the product. That is why I never consider a product is a success until it is a success.
My understanding is one more failure can be detrimental. I had invested in brn with high risk reward ratio and will wait longer than anyone else. But got frustrated when management keep on changing wordings.

So you are saying, Akida 1000 has no profit (revenue) and blaming on management.
We all know our main revenue stream will come from MegaChips and Renesas in royalty form.

Does it mean they have bad products and sales team that nobody wants.

Please correct me.

Or Let's hire this guy

 

[Diogenese]

IBM AGI patent - The Mechanical Turk

https://upload.wikimedia.org/wikipedia/commons/6/6e/Racknitz_-_The_Turk_3.jpg

US2023135625A1 AUTOMATED GENERATION OF DIALOGUE FLOW FROM DOCUMENTS 20211028

A computerized method, system and computer program product for building a dialogue flow. One embodiment of the method may comprise receiving an input document, the input document comprising content, and generating, by a question-answer pipeline, a plurality of question-answer pairs from the content of the input document. For each question-answer pair, the method may further comprise feeding the question of the question-answer pair into an intent of a dialogue flow structure, and feeding the answer of the question-answer pair as one response of the intent. The method may further comprise tagging each of the plurality of question-answer pairs with a corresponding document section index, reading, by a conversational agent, the input document to a user, pausing the reading when the conversational agent reaches one of the document section indices in the input document, and in response, reading the question corresponding to the document section indicia to the user.

 

[mrgds]

So you are saying, Akida 1000 has no profit (revenue) and blaming on management.
We all know our main revenue stream will come from MegaChips and Renesas in royalty form.

Does it mean they have bad products and sales team that nobody wants.

Please correct me.

Or Let's hire this guy
View attachment 43177

i always thought there was a connection between Shawn Hehir & Fact Finder ..............

 

[Pappagallo]

It definitely is. I've also exhausted my funds and deeply red The opportunity cost of being in red also frustrates me. I could have made an easy $10k on 4DS but decided to put it into Brainchip at 32cents.
4DS is not even making revenue yet it's booming. Something seriously wrong here!!!

This time last year 4DS were almost read the last rites. People were waiting to see if they could even physically produce their technology after two failures. It would’ve been brave to load up on them IMO.

We’re in a similar position though in the sense that people are waiting for something, but it’s not the technology because that has been proven. People are waiting for sales, which is not a given but there’s certainly less risk involved than a technical validation. But when the time come there’s every chance we could experience the extreme turnaround that 4DS is enjoying now.

 

[Frank Zappa]

So you are saying, Akida 1000 has no profit (revenue) and blaming on management.
We all know our main revenue stream will come from MegaChips and Renesas in royalty form.

Does it mean they have bad products and sales team that nobody wants.

Please correct me.

Or Let's hire this guy
View attachment 43177

Is that Fact Finders Son ??

 

[Rickk]

It definitely is. I've also exhausted my funds and deeply red The opportunity cost of being in red also frustrates me. I could have made an easy $10k on 4DS but decided to put it into Brainchip at 32cents.
4DS is not even making revenue yet it's booming. Something seriously wrong here!!!

Shorters game!!! just look at the top shorted stock chart on ASX, BRN already reached top 5 7 days ago. The wrestling between the believers and shorters are getting more and more fierce. In the heat of battle, as victory or defeat hangs in the balance, the cruelty of conflict often reaches its zenith. But one thing is clear, shorted shares need to be returned to the owners one day and their profit is always limited.

32c is still a good buy. Hang in there folks!