BRN Discussion Ongoing


Interesting video and only a couple of minutes long.

They have reduced the labelling of 1 hour of driving video to 50 hours which is very impressive until you realise that Tesla has somewhere in the order of 3 billion miles of driving video.

To make it easy this means for every billion miles travelled at 100 kph you would generate 10 million hours of video which times 50 hours becomes 500 million hours of labelling that needs to take place. At 3 billion miles this becomes 1.5 billion hours. It is easy to see why some think this is unsustainable.

What we really need is one shot and incremental learning on device I wish someone would invent it.

If the maths is wrong it was Blind Freddie otherwise
My opinion only DYOR
FF

AKIDA BALLISTA
 
  • Like
  • Haha
  • Love
Reactions: 39 users

Proga

Regular
Rob also later mentions “over 5000 unique users” of the MetaTF software. Pretty sure it was 4500 only a few weeks back. So what does this increase of 500+ users represent? New companies? If so, how many? 5, 10, 50, 100? Or is it existing companies ramping up their development? Maybe a bit of both?
I'm assuming unique users means new companies. It's a validation BRN's new marketing team and strategy is working. As others have previously commented, BRN's product is becoming well know globally and potential end users are now exploring how they might integrate Akida into their products. It's now up to the sales department to go to work.

Rob did say this may take months if not years to get a product to market. Every company is unique in regards to

1/ how fast they're able to comprehend the new tech
2/ where they're at in their product lifecycle
3/ what level of risk they're prepared to undertake
4/ the size and budget of their R&D teams

As Rob touched on, the great thing about the developing ecosystem is other companies (Megachips, Edge Impulse etc) are positioning themselves to be able to help develop applications no matter where each company is positioned within the edge AI spectrum (beginners to advanced with massive R&D teams). All you need is an idea and the sales team can either help internally or point you in the right direction at one of BRN's partners to get you started.
 
  • Like
  • Fire
  • Love
Reactions: 25 users

Slymeat

Move on, nothing to see.
This is from the datasheet of the product:

Analog interfaces
□ Ultra-Low Power Voice Activity
Detection (VAD) enabling seamless
audio processing with system-on
current < 26 uA
. . .
@uiux I feel I have to explain my 🤣 response to your reply. I believe I understand your response, and as an engineer, appreciate the way it is delivered. Considering that µA and max mV are involved is quite telling, but without definitively knowing, you can‘t just give a simple ”yes” answer. Analogue interface is a bit of a firfy but nothing an ADC can’t fix.

That power consumption seems to scream spiking to me.

Seamless also implies minimal lag, so all logic must be self-contained.
 
  • Like
  • Fire
  • Love
Reactions: 22 users
Not sure if this has been mentioned but all this is coming up:

Sensors Converge Exhibition and Trade Show in San Jose, Calif., June 27-29


A lot of interesting things happening here.

There are also co-expos happening as well:

- Embedded Technologies Expo & Conference, the largest embedded and IoT gathering in North America and taking place June 27-29 in San Jose, CA
- Autonomous Technologies Conference, June 27-28 in San Jose, CA
- Metaverse Global Congress, June 28-29 in San Jose, CA
 
  • Like
  • Love
  • Wow
Reactions: 27 users

Slymeat

Move on, nothing to see.
Just received this newsletter via email.https://www.edge-ai-vision.com/latest-news/newsletter/?utm_campaign=EVI%20Newsletter&utm_medium=email&_hsmi=217201181&_hsenc=p2ANqtz-9mt67SVbR4NNylhbRO7T1MIdAgPtSDHuR_ukxvAehjBEhwTauuqXh6-eYrwQ-UzyewdWnCw5HjVDXLLSZZ6DnuDxUseA&utm_content=217082579&utm_source=hs_email
This is the website page if it makes it easier?

To view this newsletter online, please click here
DEVELOPING SECURE VISION-BASED DESIGNS

IoT and Vision: Why It’s a Security Minefield and How to Navigate ItArm
Recent advancements in machine learning have enabled market innovators to build insights from IoT sensors in the wild. These insights can be . . .
I wish people would stop using IoT when the processing is done on the edge and doesn’t involve the internet at all. The term AIoT (as I recently updated it on Wiki) is much more meaningful—Artificial Intelligence of Things—as in no internet involved!

And as is quite well stated in the article shared by @Boab, the IoT is a mess were privacy is concerned. The AIoT has no such issues!!!

Back in March this year I added the following paragraph to the wiki page for AIoT to make it more pertinent to Akida, and more difficult for others to say they are using it:

”In the AIoT an important facet is AI being done on some Thing. In its purest form this involves performing the AI on the device, i.e. at the edge or Edge Computing, with no need for external connections. There is no need for an Internet in AIoT, it is an evolution of the concept of the IoT and that is where the comparison ends.”
 
  • Like
  • Love
  • Fire
Reactions: 34 users
Hi DB.
As i mentioned in yesterdays post about the Shorting , wait to see if they (THM/HC/MF) have done more Shorts on Tuesday (yesterday) as the trading pattern seemed to suggest they had.
Todays action shows more "Shorting" activity thought it is very small really considering the weak market (ASX) today.

Well the Data is in and ............Yes they did take out some more "Shorts" yesterday - 3.8 Mil NEW shorts for Tuesday
after Mondays 3.3 Mil and Fridays 11.3 Mil.
Quite a lot of shorts................and the SP is still up around 0.90c aprox.

NOT going very good for them so far! lol

Yak52 :cool:
Well you know Yak, every lot of borrowed shares, sold on market, is like adding another piece of wood to the pile.

They must buy them back at some stage and if they get caught out, with a significant development from the Company...
Say an IP deal with Moschip, or a partnership, with the likes of Nvidia (neither of which are fanciful).

Well then, you don't want to be anywhere near it, when that goes up! 😛 🔥🔥🔥

_20220622_143700.JPG
 
  • Like
  • Fire
  • Love
Reactions: 36 users

Slymeat

Move on, nothing to see.
The old double edged sword trick. Need to use more coal to cool down, so makes it warmer, so need to use even more coal!
China have it all sorted, their LFTRs will be coming on line very shortly. Actually their first test plant is supposed to be online already.

China has plans for 40 LFTRs, that will give them essentially unlimited clean energy, plus all the you-best byproduct isotopes that are necessary in medical circles and not able to be produced by any other means.

And historically, with abundant power comes power - as in geo-political power. This scares me more than anything else at the moment.

And to think the FISRT nuclear reactors were LFTRs. And that was in the 50’s by a team of Australian scientists. LFTRs were abandoned because they didn‘t produce plutonium/
 
  • Like
  • Fire
  • Love
Reactions: 10 users

If you are wondering about whether AKIDA might be part of Nviso’s medical engagement with Siemens Medical the following is on point in particular the very last paragraph:

Tim Llewellynn

CEO/Co-Founder of NVISO Human Behaviour AI | President Bonseyes Community Association | Coordinator Bonseyes AI Marketplace | IBM Beacon Award Winner #metaverse #edgeai #decentralizedai
7h

Wow ResApp Health, a startup in Australia commercializing an AI App on the mobile phone to screen and diagnose cough sounds, just lost A$53m of value in it's buyout from Pfizer because it's AI App couldn't reproduce trial results to a required degree of accuracy. If you are a #shareholder of an AI listed company there are some learnings in this! At NVISO we have some experience in what it takes to bring clinical and high performance AI Apps to market working with the likes of Siemens Healthineers and med-tech research partners Innosuisse, Technical University of Munich, and European Commission for over a decade now. Here are a few learnings based on our decade of experience: - Data-driven AI technology is not a buy and forget technology. It requires constant data and feedback to improve. So that you can maximize providing quality care to everyone in the world - there is really no such thing as an "outlier" in medical AI applications. - While data is king, an automated, dependable, repeatable and well maintained data-driven software tech stack to constantly train, test, deploy, verify, and explain the AI Apps during dev/prod is the queen, bishop, knight, and rook combined. - When training your AI Apps on medical data don't go cheap on data collection and labelling. Collect more data than you need, ensure you have enough samples across the population with noise factors included, ensure you cover variance across ethnics groups, skin colors, age, gender, disabilities, and try to get towards a 99% population coverage as early as possible. - Don't outsource the entire data collection and labelling to a university - you learn a lot by working directly with the data yourself, understanding the labelling process, etc - To ensure robust generalizable performance performance with small datasets, ensure that datasets and models are constantly upgraded and using SOA in terms of best practice (this can change every 6 months in the field of AI so constantly review SOA publications every month). Don't focus solely on model architecture and training a lot of great improvements can come from data-centric AI. - Set aside appropriate development budgets to constantly invest into an "AI Flywheel" -> users -> generate data -> build better AI model -> acquire more users -> etc - This allows you to constantly improve your AI App during development cycles and improving generalization performance. - Ensure you have separate training, test and verification datasets. Verification datasets MUST come from a source completely independent from your training dataset. -

Ensure you have the right team and infrastructure in place to build a core team or if you can't find the talent engage a specialized AI company to assist you with the journey”

Remember Brainchip is described by Mercedes Benz as artificial intelligence experts and they have allowed Nviso to achieve WOW performance levels.

My opinion only DYOR
FF

AKIDA BALLISTA
 
  • Like
  • Fire
  • Love
Reactions: 41 users

Reuben

Founding Member
Great Sony Video


Imaging & Sensing​

Unlocking the potential of imaging and sensing technology with AI to unveil new opportunities

Our Approach​

We are working on combining industry leading sensor technologies developed at Sony Semiconductor Solutions with new machine learning methods and robotic actuators. By developing novel learning and control algorithms, while also evolving the computing hardware to handle the sensor data in optimal ways, we will discover and explore the full potential of Sony’s sensor technologies and sensor-based solutions.We believe that the resulting systems will open up applications for imaging and sensing technology in areas that no one has imagined before.



@TechGirl thankyou for this video, here is another Sony Incabin video.

 
  • Like
  • Fire
  • Wow
Reactions: 26 users

TechGirl

Founding Member
@TechGirl thankyou for this video, here is another Sony Incabin video.



I like it & I also like that it is Sony demonstrating in a Mercedes, I can't remember do we have a connection with Mercedes :ROFLMAO::ROFLMAO::ROFLMAO:
 
  • Like
  • Fire
  • Love
Reactions: 46 users

Dang Son

Regular
Hi DB.
As i mentioned in yesterdays post about the Shorting , wait to see if they (THM/HC/MF) have done more Shorts on Tuesday (yesterday) as the trading pattern seemed to suggest they had.
Todays action shows more "Shorting" activity thought it is very small really considering the weak market (ASX) today.

Well the Data is in and ............Yes they did take out some more "Shorts" yesterday - 3.8 Mil NEW shorts for Tuesday
after Mondays 3.3 Mil and Fridays 11.3 Mil.
Quite a lot of shorts................and the SP is still up around 0.90c aprox.

NOT going very good for them so far! lol

Yak52 :cool:
Hi Yak, It's not just BRN .
These Multi National Insto Bots seem to be shortING every promising Aust Co.
See FMG WBT DEG AXE GRR etC
Legislators should put a bomb under them . They generate brokerage and GSt on that is about all , while stripping massive equity out of the Country
 
  • Like
  • Fire
  • Love
Reactions: 34 users
A. Akida
S. Standard
S. in plain Sight
U. Ubiquitous
A. AI
G. Game changing
E. Essential/Edge
D Defacto

Akida ASSUAGED the need for ultra low power at the edge.

Very clever @Slymeat Rearranged to suit your name

S piking
A kida
U biquitous
S in plain Sight
A I
G ame changing
E ssential
 
Last edited:
  • Haha
  • Like
Reactions: 16 users
I like it & I also like that it is Sony demonstrating in a Mercedes, I can't remember do we have a connection with Mercedes :ROFLMAO::ROFLMAO::ROFLMAO:
I’m missing the balance of the other place so yes but it it is only with Mercedes’ Benz luxury passenger vehicles not Mercedes’ Benz luxury Commercial.😂🤣😂

FF

AKIDA BALLISTA
 
  • Haha
  • Like
Reactions: 10 users

Esq.111

Fascinatingly Intuitive.
Don’t forget the NVISO webinar tomorrow, they will be doing a live demo using Akida and will also be taking questions, so get your questions ready.

It’s at 7pm Sydney time (AEST)
Here’s the link for those who haven’t registered-

Afternoon Chapman89,

Hope you don't mind, thought I would drag your post to the front of the queue.

Thankyou .

Looking foward to watching.

Regards,
Esq.
 
  • Like
  • Love
  • Fire
Reactions: 19 users
Hi FF, my minds in a bit of a quandary in regard to Intel's ( Loihi ) and IBM's ( TrueNorth ) and there usefulness only as research hardware ?. Are they both still being developed further, and could still be contenders, ( competition ) as they are or could be, ??.

For years it has been stated that these neuromorphic chips won't have any commercial value to fabricate, and sell and are only useful for research purposes. As one notable professor often muttered, ( why is it so ? )

Wouldn't Akida itself be more than useful as a research candidate ( because it's more powerful etc ) than the Intel or IBM offering, particularly as it has much more features, ( useful features ) and is far and away considered more ( state of the art, so to speak ) leading edge than the other 2 offerings.

Also, because the Loihi and TrueNorth processors won't ever be considered worthy of commercialization, ( and only worthy for research purposes ) doesn't this mean that Akida is the front runner both for research and as a commercial ( in the " first to market" sense ) solution.

I must be missing something, the boss ( i.e. she who must be obeyed ) often picks me up on the somethings in life that we come across and the missings of same. ;) It's not unusual. ( to be loved my anyone )

Just thought I'd put this concern ( not a worrying one, hopefully ) out there ( and is not a daft query ) for some comment from all/any of the technical wizkids ( that appear to be in abundance, on these threads ). I'll just sleep better for having this quandary removed from my thought processes for good.

Akida Ballista >>>>> Are we getting there driver ?? and, How soon will we arrive <<<<<

gltah and followers ......


hotty...
Awesome response and history from FF on this question posed above. Also to add on to this conversation and this is my experience and perspective.... if you are a IT hardware/IP developer you don’t want to be just used as research hardware.

This is a general clear sign that the tech is “not fully working as we would like yet” and “still in development”...ie “its got some bugs”

I come from the ICT world and the view to me of a hardware vendor just placing their new tech out in the world for “research use” screams that this tech is really just at “proof of concept” stage and they don’t guarantee anything about it as yet. So they are saying use it (most likely for free) and have fun but it may have problems and we don’t really support it fully but let us know about the problems so we can then fix them all as we go.

Brainchip have achieved “fully tested” and “production” status and to achieve that for bleeding edge AI tech is very difficult and time consuming(as most of us know now) and It now places Akida in a whole other league.

So it is great to get into universities and research groups now for Brainchip (while not generating much dollars) this will help generate the value of being an “industry standard” or “best practice” and is available for production/commercial use is one of the biggest benefits.

Intel and IBM have not yet achieved what Brainchip has as their chips are not finalised/solved to even start many rounds of “production testing” and has not achieved the final state of “production ready” status so it can be released to the world for real world commercial use.

Would a Uni prefer to study a theoretic product or a real world commercial product.

Saturation of Brainchips products are occurring now....how many NDA’s :) ...how many high profile names associated now or broken cover to basically say there is a best practice hardware to use for Neuromorphic computing ....it’s Brainchip. This to me means the Uni’s will literally be chasing down Brainchip products for use in classes over time as their is no substitute for the smarts or anything else in true production that comes close.
 
  • Like
  • Love
  • Fire
Reactions: 42 users
D

Deleted member 118

Guest
  • Like
  • Haha
  • Fire
Reactions: 20 users

Pokitren

Emerged
Awesome response and history from FF on this question posed above. Also to add on to this conversation and this is my experience and perspective.... if you are a IT hardware/IP developer you don’t want to be just used as research hardware.

This is a general clear sign that the tech is “not fully working as we would like yet” and “still in development”...ie “its got some bugs”

I come from the ICT world and the view to me of a hardware vendor just placing their new tech out in the world for “research use” screams that this tech is really just at “proof of concept” stage and they don’t guarantee anything about it as yet. So they are saying use it (most likely for free) and have fun but it may have problems and we don’t really support it fully but let us know about the problems so we can then fix them all as we go.

Brainchip have achieved “fully tested” and “production” status and to achieve that for bleeding edge AI tech is very difficult and time consuming(as most of us know now) and It now places Akida in a whole other league.

So it is great to get into universities and research groups now for Brainchip (while not generating much dollars) this will help generate the value of being an “industry standard” or “best practice” and is available for production/commercial use is one of the biggest benefits.

Intel and IBM have not yet achieved what Brainchip has as their chips are not finalised/solved to even start many rounds of “production testing” and has not achieved the final state of “production ready” status so it can be released to the world for real world commercial use.

Would a Uni prefer to study a theoretic product or a real world commercial product. I will also recommend https://betpokies.com, where you can find great reviews of casinos in Australia.

Saturation of Brainchips products are occurring now....how many NDA’s :) ...how many high profile names associated now or broken cover to basically say there is a best practice hardware to use for Neuromorphic computing ....it’s Brainchip. This to me means the Uni’s will literally be chasing down Brainchip products for use in classes over time as their is no substitute for the smarts or anything else in true production that comes close.
Hi! I support what you said about Brainchip - they look promising.
You also talk about companies that distribute their product for free, but collect feedback about problems. Do you think this approach just saves the product budget?
 
Last edited:

buena suerte :-)

BOB Bank of Brainchip

A little write up from..... STOCKS down under ..​

How to benefit from BrainChip’s breathtaking potential in 2022​

Behzad Golmohammadi Behzad Golmohammadi, June 17, 2022

Who is BrainChip?

BrainChip Holdings Ltd (ASX: BRN) develops software and hardware solutions for artificial intelligence (AI) and machine learning applications. The company’s key product is the Akida Neuromorphic Processor, which is in essence a Spiking Neural Network (SNN). It provides ultra-low power and fast AI (artificial intelligence) Edge computing solutions without the need for a continuous internet connection. Additionally, unsupervised learning capabilities and on-chip processing (rather than in the Cloud) set Akida apart from other AI solutions on the market.
Akida can be used in a range of applications, like vision and audio, which in turn are used in a wide variety of industries, such as automotive, robotics, aerospace and cybersecurity to name just a few. BrainChip was incorporated in 2011 and is based in Sydney, Australia.

We have covered BrainChip and its revolutionary SNN technology extensively at Stocks Down Under and Pitt Street Research. You can use the following links for an in-depth explanation of BrainChip’s technology and business model:

Additionally, we have also written about BrainChip multiple times in Stocks Down Under, accessible to Stocks Down Under subscribers.

In this article we’ll look at what is driving BrainChip’s share price and how you can play this high-potential stock in a volatile share market from a technical analysis point of view. Let’s start by taking a look at BrainChip’s share price chart to see what moved the share price after the Corona Crash.



BrainChip

BrainChip Holdings, Weekly Chart in Semi-log Scale (Source: Metastock)


❶ BrainChip announces an agreement with a US-based automotive manufacturer for evaluation of BRN’s Akida neural network System-on-Chip (SoC) for Advanced Driver Assistance Systems (ADAS) and Autonomous Vehicle (AV) applications. (Akida Evaluation Agreement)

❷ BRN signs a joint development agreement with a tier-1 European automotive supplier of sensors and systems to utilise Akida for ADAS and AV applications. (Joint Development Agreement Executed)

❸ BRN announces completion of wafer fabrication of the Akida device to be used in the evaluation programs of those customers who signed agreements with BrainChip. (Akida Wafer Fabrication Complete)

❹ BRN announces a partnership with Magic Eye Inc., a developer of revolutionary 3D sensors, to use BRN’s Akida neuromorphic processor in object detection applications. (Partnership with Magik Eye Inc)

❺ BRN signs a collaboration agreement with VORAGO Technologies, a US-based technology company, to support a Phase 1 NASA program for a neuromorphic processor that meets space flight requirements. (Vorago Technologies Collaboration)

❻ NASA places an order under the Akida Early Access Program and BRN signs its first Akida IP license agreement with Renesas Electronics America Inc., a tier 1 semiconductor manufacturer. (NASA Early Access Program Order)(First Akida IP License Agreement)

❼ BRN addition to the ASX 300 Index. (S&P DJI Announces March 2021 Quarterly Rebalance)

❽ BRN begins taking orders for the Akida AI Processor Development Kits. (Taking Orders for Akida AI)

❾ BRN enters into a multi-year license agreement for the Akida IP with MegaChips, a pioneer in Application Specific Integrated Circuits (ASIC’s), to enhance its technology positioning for next-generation AI solutions. (Megachips’ License Agreement)

❿ Increased media attention to the AI industry and tier 1 companies joining BRN’s Early Access Program customer list increases retail investors’ attention and confidence in the stock. (Response to ASX Price Query)



ADAS and autonomous driving are major catalysts for BrainChip

The COVID-19 pandemic introduced unprecedented challenges to the semiconductor industry, which led to global chip shortages that automatically brought substantial investor and media attention to stocks in this industry. In addition, the near-zero interest rate environment encouraged investment in growth and speculative tech stocks. At the same time, there have been ongoing discussions surrounding climate change and how to reduce global carbon emissions, which has further propelled the Electric Vehicles (EV) revolution.

An important point of competition amongst EV manufacturers is their vehicles’ driver assistance and autonomous driving capabilities. While Tesla, the American automotive and clean energy company, is way ahead of its competitors in terms of autonomous driving and driver assistance capabilities, traditional auto makers getting into EVs are striving for any technology that can help them in their battle with Tesla. This is why BrainChip’s Akida neural network System-on-Chip (SoC) is receiving increasing attention from tier 1 auto and auto parts makers in Europe and elsewhere.

Looking for advanced technologies to fill their knowledge gaps, they have come across BrainChip that checks all their ADAS boxes when it comes to required semiconductor IP. No wonder, then, that every time the company announced new milestones and deals, the share price reacted very sharply, to the extent that it grew from 3 cents in March 2020 to $2.34 in January 2022, an astronomical growth of 7700% in less than two years.



Less and less risky, but with high potential

After BRN’s exponential growth through January 2022, the share price started a correction in line with the rest of the tech sector as the enthusiasm for speculative tech stocks started to wane due to fast rising interest rates. This share price decline, however, might not be all bad news for the true believers in BrainChip’s technology as it presents new opportunities to buy the stock at lower levels.

Many of the reasons that made BrainChip attractive in the tech boom are still in place. There are still chip shortages and car manufacturers still need the relevant semiconductor IP and the actual chips to give them an advantage over their competitors, or to play a role in ADAS and AV in the first place.

And BrainChip’s technology has kept showing more and more promise by successfully passing product development milestones. There is now a number of tier 1 companies testing the utility of BrainChip’s Akida AI processor kits for their respective applications, which increases the chances of getting some good news flow any time in the following months.

In other words, we believe BRN has largely completed its development stage and is now moving into the commercialisation stage under the leadership of new CEO Sean Hehir, who we spoke to recently. As a consequence, we believe BRN, the share, is getting less and less risky.

On top of that, BRN has just been admitted to the ASX 200, which will trigger mandatory institutional buying as BRN now falls within the investment mandate of many institutional investors that invest in the ASX 200.



Upside potential to $2.30

As the interest rate hikes by central banks and the sell-down of risk assets continue for the foreseeable future and in the potential absence of any exciting news from BRN in the near term, we expect the share price to drift lower under its own weight.

However, in a more likely scenario, at least in our view, there will be successful development news from one or more of BRN’s Early Access Program customers or news around new commercial deals with tier 1 or tier 2 semiconductor companies. Given that the company is in discussions with many prospects at the same time, such news flow can be expected at any time and could have the potential to send the share price back up towards its all-time high of $2.34 in the initial market reaction to the news.
 
  • Like
  • Fire
  • Love
Reactions: 50 users
webinar is starting
 
  • Like
Reactions: 9 users
Hi DB.
As i mentioned in yesterdays post about the Shorting , wait to see if they (THM/HC/MF) have done more Shorts on Tuesday (yesterday) as the trading pattern seemed to suggest they had.
Todays action shows more "Shorting" activity thought it is very small really considering the weak market (ASX) today.

Well the Data is in and ............Yes they did take out some more "Shorts" yesterday - 3.8 Mil NEW shorts for Tuesday
after Mondays 3.3 Mil and Fridays 11.3 Mil.
Quite a lot of shorts................and the SP is still up around 0.90c aprox.

NOT going very good for them so far! lol

Yak52 :cool:
"Well the Data is in and ............Yes they did take out some more "Shorts" yesterday - 3.8 Mil NEW shorts for Tuesday
after Mondays 3.3 Mil and Fridays 11.3 Mil.
Quite a lot of shorts...............
.and the SP is still up around 0.90c aprox"

II just had an epithany Yak52..

It looks to me, like they are doubling down on their bets against Brainchip.

They could be in serious trouble here!

And I mean serious!

🔥🔥🔥
 
  • Like
  • Fire
  • Haha
Reactions: 28 users
Top Bottom