Skip to content

Dynex AMA #002

Friday, March 31st, 2023 | 14:00 UTC


Dynex AMA #001

Saturday, January 14th, 2023 | 13:00 UTC.

Transcript AMA #001

Hello everyone and welcome to our very first AMA! We are very excited to have all of you tuning in. Today, we have Sumitomo the founding father and Mastermind of DYNEX joining us to talk about the vision of the project future development, technical aspects, and answer our questions. Pleasure to have you here with us, Sumitomo.

Hi everyone, this is Sumitomo here. I’m wondering if everyone can hear me?

I think, everything is fine, I can hear you, most likely we all can hear you. Thank you very much!

So, my name is Cara, and I will be moderating and guiding today’s session. The AMA will consist of two rounds. The first round will consist of the introduction to DYNEX, and then, the second round, will be us going through the questions that I had prepared and collected through our social media channels for you beforehand. We will also keep an eye, in the meantime, on the incoming questions on our Discord channel. So, feel free to write your additional questions. I’m keeping track of them. So, let’s dive in! 

Dear Sumitomo, can you please introduce yourself, as well, as tell us about the developer team behind DYNEX?

Well, at this point, as you know, our identity has not been disclosed, but what I can tell you is, that we are very experienced team in building successful tech companies, many of those you know very well, and you actually, probably, using them in your everyday life. The team consists currently of about 25 members, and when we started the project, we did it as an attempt to create impact on our society by actually solving actual real-world problems and problems that until now have not been solved. The whole project is fully self-funded, and it strongly relies on participating miners, because we are building a neuromorphic network here, together. Our vision is to make the world more sustainable and use the resources to maximum potential and impact. 

Wow! This was a very interesting introduction! Rather big statements, I must say. Would it be so kind to share the backstory of DYNEX?

Absolutely, so, I think everyone is aware of the fact that we are in the midst of the AI era. The global Artificial Intelligence market is growing rapidly. I mean, you see numbers here, but without diving into what is actually common sense. I want to give you a little bit of background first. So, AI is growing we know that, but there’s also challenges coming with it. So, we have an exponentially growing computing demand, so, the more people are applying AI, the more computing power we need, and that actually leads to a couple of challenges. Challenge number one is: we have actually not created another faster device in the same speed as it as it used to be, because we cannot make transistors smaller and smaller and smaller. So, basically we are running into a wall here, right? And if you look at it also from an architectural development standpoint, even that performance gain is actually slowing down as well. So, if you look at it from a technology perspective, and also from an architectural perspective, we are running into the so-called “von Newman bottleneck”, without explaining it further what exactly is it. So, all existing computers, as we know them today are basically bound by the communication throttle, which is the data bus, right? So, we have CPU, we have memory, and these two need to communicate with each other. That creates a memory of all, and a power wall. In fact, if you look at it right, we actually have created really powerful things. Think about AlphaGo, for example, right? Manage to beat humans in goal, but at what cost, right? We are talking about 150 000 Watts, right? Our human brain is so much more efficient! We are actually talking about 20 watts, so, what we see is, there is an urgent demand for new devices and new architectures.

So, practically what you are saying is that our current way of computing is inefficient to make new substantial development. It is unsustainable in the long run. Do I understand it correctly? And what I would like to know is, what is the actual solution you’re proposing?

So, what we found extremely interesting and which is actually a discovery. Is an invention or a discovery, from a gentleman called Leon Chua. In 1971, he created a theoretical model, of what he called the missing circuit element. So, if you step back a little bit, if you look at current circuits, they all consist of capacitors resistors, and, inductors. What he created, and it was later on, also being picked up by Hewlett Packard, is basically what we call the memristor. What is a memristor? In simple terms, it is a transistor which also contains memory, and, as it contains memory you can actually very, very, accurately simulate real Neuromorphic Computing, similar to what’s happening in our brain. So, that’s basically called: in- memory computing.

Oh, wow! That was definitely extremely interesting, but a little theoretical though. Can you please explain, how is all this connected to DYNEX?

Sure, so let’s take a neuromorphic chip which is a theoretical model, right? What we did is – we basically created a logic gate, which does not consist of transistors, but it consists of memristors. What does this mean? These logic gates do not have to use any bus structure. They don’t have to communicate with any memory. So, they are parallel. They provide in-memory Computing, and they are highly efficient at low power. So, if you take any computational task, and we listed them here on the screen. We actually take that computational task, and we convert it into a circuit, and this circuit is basically built up with our neuromorphic chips. Now, how what are we doing it, right? These neuromorphic chips are run on the miner’s computers. That’s the algorithm called DynexSolve, and it’s a Proof of Useful Work algorithm, obviously, right? Because what miners are doing, they are taking this gigantic neuromorphic chip circuit design, and every participating miner is computing a small piece of it, and DynexSolve has been created completely preparatory. So, it was launched on December 1st, and it allows us to perform computations in a distributed manner, in a malleable environment, whereas the malleable environment is basically an environment what we see in mining, because miners come and go as they want. They restart it, it grows, it becomes bigger, it becomes smaller. It’s kind of challenging, but we managed to find a way to build a blockchain in order to facilitate it. 

So, just to summarize, so that we understand you correctly. That every miner is contributing to the neuromorphic network, and while mining it actually is solving a part of computational task. Is that right?

Absolutely correct.

Okay, that’s great! I think we are on the same page here. Do you have any examples of these computational tasks and problems, so we can understand what these are and what are we actually solving?

Absolutely, I prepared a few. So, one very interesting example and that’s actually one problem our network solved already, is how to optimize the energy consumption, which is especially important for electrical vehicles. Because, if you want to go from A to B, you want to spend as much less energy as possible into getting there. So, by applying our technology we managed to lower the energy consumption from 86 kilowatt down to 3.8 kilowatt, and this is significant! 

Another example is, obviously, if we can improve machine learning, right? Anything around image recognition, medical diagnosis, anything AI is being applied to. The accuracy levels go from 76 percent to 97, which means we can do machine learning tasks faster and better, and maybe more graspable. For example, protein folding problems, most of you have heard about it. What is it? It basically allows companies to discover drugs faster, because they can simulate protein folding, they can try new things, and the recent advances for example in COVID medicine, was all based in protein folding problems. And lastly, obviously, and this is personally for all of us, really an important topic, we can finally leverage AI much more efficiently, because we need environmental sustainability for all problems associated.

Oh, that is definitely very, very, intriguing! I think I’m not the only one who thought about Corona when you spoke about protein folding and new drug finding. That definitely sounds extremely impressive. I think, a lot of attention will be and should be paid towards the sustainability goals. As we know that this is actually one of very large issues these days. All the countries are trying to resolve them and find solutions, so this is definitely something that makes us very inspired. Thank you very much! I do have another question though. So, I would like to know if are there any other technologies or computers existing that can also solve these problems? We all know about SAT competitions, for example, where people from all over the world are participating with their solvers and are resolving some of those some of these problems. So, please tell me how is DynexSolve different?

Excellent question and we actually did performance tests and comparisons with exactly SAT competition participants and solvers. To give you a couple of examples which I have prepared here, for example there is an NP heart benchmark problem. What does this mean? It’s basically a problem which is very complex, very hard to solve for regular technology because it’s based on ferromagnetic phase transitions. I don’t want to get into the details, but what I can tell you is – if you take the winner of the international SAT competition actually, even 2021, doesn’t matter, or if you take any commercial product – it can simply not solve this kind of problems. When you reach a certain level of complexity. With DynexSolve, we solve that very, very, efficiently, and faster than any other solver out there. Or a second example, which is also quite interesting. In these SAT competitions, for example, if you take, let’s say 300 randomly chosen SAT problems and then you take also the winners of the international SAT competition, they solve around 60 percent of those, in a time frame of around five hours. These are very simple problems in reality. If you run those through DynexSolve you solve a 100 percent in 30 seconds.

These are definitely some impressive numbers! Okay, why don’t you tell us a little bit about your business model?

Yeah, well, business model is also kind of important, isn’t it?

So, basically, I’ve prepared one slide outlining the Dynex ecosystem and come in combination with the business model. So, the way it works, first of all we are in an exponentially growing market demand, so customers will need more and more, and more computing power, and they also want as less energy being spent on that as possible. So, what are these? Corporates, organizations research bodies, practically anyone, who is in that in that field, and who basically applies artificial intelligence or logistic problems, optimization problems. Whatever it is, is basically a potential customer. Now, in order to equalize the pricing, we have created what we call, a demand driven pricing model, which is the Dynex Marketplace. So, these customers can come to the Dynex Marketplace and post their job offers, so to say. And they define – okay I’m going to pay that amount of block fee, I’m going to pay that amount of solution reward. The miners on the other side, they pick obviously the best paying jobs first, so they are prioritized. Dynex developers will get a percentage of these block fees and obviously, of the solution rewards because eventually we also want to build a sustainable business model for us. Beautiful thing here is that this guarantees, and this is very unusual for coin projects, we have a sustainable long-term model. Because even if all blocks have been mined, there will be still not only transaction fees, but block fees paid from the customers to the miners, which makes this really sustainable in the long-term.

This makes us miners very, very curious and very excited about this. This sounds amazing! We would like to know what are your future prospects and plans for DYNEX or what can we expect? Is there anything new coming up?

Well, you know we have our roadmap and I’m going to bring that up here right now. So far, we have been trying to be very punctual and accurate with our planning. I’ve seen some questions, you know what our roadmap is beyond ‘24 ‘25 ’26, maybe up even to 2030. The reality that matters is, it’s actually very hard to predict what’s going to happen in five years. So, we prefer more an agile approach, where we can basically time things we are doing here in a proper way, and actually also deliver on it. So, the biggest question of everyone is secondary listings. Yes, we’re talking about it. We have been talking about it. We are working on it, and the goal is to actually at least one secondary listing during January. But there will be more, because we see if that grows, what we have been seeing, what we are expecting, us, actually on a number of exchanges out there. Another important part is, obviously the Ethics Committee, which is basically the gatekeeper of problems out there. Because, you can imagine, if we start really building up this network into a large scale, this technology potentially could be misused as well. So, there will be an Ethics Committee, and we’re going to get to this a little bit later. And then, obviously as well, we’re going to start heavily going into marketing and promotion as soon as we have the secondary listings mapped out. 

That’s very, very nice. Great! We are definitely very curious who will be on the Ethics Committee and would definitely like to hear more about it. I would say, that was a great introduction. Thank you very much and let’s proceed to our second round where I will be asking a few questions posted by our community. I can also see some questions coming in our Discord channel in our #AMA section. Let’s start! One of the community members asked the question, if also old computers or even mobile phones can be used for mining or is it only a GPU play?

Well, obviously GPU, massively parallel, has a lot of advantages. The way we’re approaching it, is basically twofold: on GPU we basically assign the initial conditions for all the simulations. That means, it is guaranteed that in N power 5, whereas N is a number of variables, we’re going to find a solution to that problem. Whoever finds it, we don’t know, because it depends on the initial conditions of the miner. Now we have done something really sweet for CPU and for that matter it would actually also work on ARM, which is basically mobile, because there obviously, we don’t have that amount of parallel tasks. You know we cannot run two and a half thousand or five thousand DYNEX chips on a CPU, but you can run four, or eight, or twelve, sixteen, whatever, right? There, we actually created random conditions. Now, there is the element of luck, that is CPU. If you let it running, it finds a solution quicker and faster than a GPU, which means you are eligible for the solution reward. Now, the solution reward, typically is in a range, which is significant current job, for example, pays a hundred thousand DYNEX for a solution reward.

Okay. Great! What steps are being done to make companies feel incentivized to buy the token in exchange for computational power owned by DYNEX, and what type of businesses would actually benefit the most from what DYNEX is offering? Do you have any examples? 

Sure! First of all, computations on DYNEX are faster and more energy efficient. Basically, you cannot really say there’s competition at this point. Customers will simply come to us, because they want to have more accuracy in the artificial intelligence models. They come to us because they want to solve problems which otherwise would take two years. They want to solve it in two weeks or three weeks, right? The other beautiful thing is: performance can be tested. We can, actually if you have a large customer coming in who’s interested, we can have him run some test problems on our Network. Let’s say, for 10 blocks, 20 blocks, 100 blocks, and then we tell him once we found the solution. We omit it, you’re going to pay for it afterwards to become a paying customer. This is an excellent model of actually teasing and proving the pudding, so to say, when they want to utilize our things. So, who are companies which are using it? I mean, you have seen the use cases before information technology companies, energy utility companies, medical pharmaceuticals, airlines, automotive, logistic transportation. Everywhere, remember, we are addressing an exponentially growing market. 

That’s really nice! Very good answer! Thank you very much! We also have a few more questions to go. What happens if there are customers that want a rather large amount of DYNEX coins in order to access this computational time? Shouldn’t the price be naturally much higher to make it more valuable and expensive to have access to such a premium computational time? What steps would be considered to make DYNEX coin more valuable and to exceed competitors? 

As I explained before, higher paid computational jobs are prioritized by the DYNEX platform. What does it mean? Customers who are paying more, will get computationally signed first. That’s basically being regulated through the marketplace, the Dynex Marketplace. Which is coming in Q3 this year probably. This supply and demand model will regulate pricing, and yes, especially post block rewards and the limited maximum supply of our coins, the value of this coin must grow significantly.

Okay, a 340-task file, it has always existed, but its content has not changed. How do you allocate new Computing tasks? The general impression is, that the computing model needs a lot of hard disk space, while the 340 only has 18 megabytes and never changes. What can you tell us about that?

That’s correct. Well spotted! The computation file is being defined by the job. now computations, It’s not that that 10 miners are working on one job and 20 miners are working on another job, customers buy entire blocks. All the miners in these blocks are basically working on the computation until it’s been solved, right? The current job which you’re working on is, for instance the 340, the DYNEX compute 340, right? Since that is solved, you will see another one. It’s been automatically downloaded, and it’ll basically be updated automatically by your miner.

Very interesting, thank you. Just another, little bit lighter question. How did the DYNEX name come up? Did you consider any other names? 

Well, yeah. You know when it comes to us engineers, we had actually a different name before, and then we consulted with the more creative people in our environment, and we got to a much catchier name. We actually like DYNEX, I mean DYNEX is strong. It expresses what we are here for, it is recognizable, and yeah, we are quite happy with that name. 

We are very happy with that name too! It’s a great name! I just wanted to jump a little back. When you mentioned the Ethics Committee, could you share a few details about structure, operation etc.?

Yes. Sure! As I said before, the Ethics Committee is going to be the gatekeeper. Because you can imagine, especially when this network grows further, there are military explanations. There are, potentially you know, blackheads possibilities here of usage. We don’t support this, that’s why we need an Ethics Committee. Especially, for the more complex tasks. These ones are going to be reviewed by the committee. The committee is currently preparing the DYNEX Manifesto. The DYNEX Manifesto will define and set the ground rules of what kinds of computations we are going to accept on the network. Because I think, we all have to have the same goal to make this world a better place and not use technology for reverse. We need to use it for the better. I mean, smaller problem instances, which are not a security risk in any case for anyone, can obviously, be automated by rule sets. But the larger ones will go through this committee and we’re going to publish our first version of this DYNEX Manifesto. Actually, in a short period of time, and we believe that everyone who is going to be part of DYNEX, from the miner, to the customer, to the defs, everyone should hold on to that Manifesto, so that we all steer that project into the right direction.

Great, I think that ethical part of any project is extremely important. What about the privacy of DYNEX compute clients and projects?

Customers are always known to DYNEX, but the customers can choose whether the information should be made public or not. When they post a job, they can say: you know what, I want to keep it private. We have a real-time view on problems, so everyone can see all the time what are we working on, what’s happening, and they can decide it, But we will always know who is the customer, who wants to compute, and what is the computational task.

Okay, got it. Do you think you could give us few insights, because obviously we are all very curious and interested, a little more information about your roadmap for ‘24 to ’26?

Yeah, we always like that question. You know, it comes from the old startup days when the people come with 10-year business plans like anyone could predict what’s going to happen in five years or 10 years. It’s kind of jokey. We prefer more short-term plans, agile development and then really, basically follow the progress.  I mean we have a very clear plan to complete the DYNEX ecosystem, and you have seen it in the world map. Everything which comes after is going to be developed as we speak. I mean there are, obviously, items which are interesting. For instance, micro jobs certain smart contracts or smart mini job features, etc. That stuff we are going to announce, but this is, obviously, too early to say now.

Okay. Great. Maybe you can tell us about: what is the current computing power of our network in the industry. This is one of the questions that one of our community members asked.

Obviously computing power typically is expressed in flops, right? Which are the floating points, operations. There are only two sides of the story. I mean, we publish our network flops on our website. Everyone can have a look at it, but more importantly, is actually the efficiency of the computation because I can spend 10 teraflops on a problem. If I’m not solving that efficiently, I will not be able of being faster. Let’s say, if you have a hundred flops, you always have to take into consideration, what are flops and what is the efficiency, in order to really compare things.

Here we have another question coming in from “Lemper”. Thank you very much! How about a block reward decreasing? He means hafing or else. Would you be so kind to comment on it? 

Yeah, absolutely! We think hafing is actually a pretty outdated model. I know that especially heritage coins still do it. We have a continuous emission schedule, so everyone who is interested how that works, simply go and read the white paper. The way it works, technically is, the block reward is basically being reduced block by block. Every block which is being mined, the next block reward will be slightly less, and you can just follow the emission curve. 

Okay, great! Thank you very much for your answer. We have another question here. Do you have any competitors? Do you see Intel as a competitor as they are coming up with a physical neuromorphic computing chip too? How do you see that?

First of all, every good project should face competition, because if there’s no competition, we would do something wrong, right? But I think there’s first mover advantage, which will allow us to stay ahead of the game. We had, so far, the one and only decentralized neuromorphic computer network, but there are other companies out there which are doing really interesting things, in terms of proof of useful work, for example. Now, that’s something we like to embrace, so our approach, in history and also with this project, has actually shown us that embracing competition is sometimes much more beneficial then fighting competition.

Thank you so much for your answer. I’m quickly having a look on the new incoming questions. Let’s see. I’ve seen that the assertion of the DYNEX chips could be used to mine other algos including presumably BTC. Is this true? Can DYNEX chips, theoretically, be used to run DYNEX of itself.

That is, in fact, an extremely smart question, because the DYNEX computation model can theoretically also solve cryptographic problems. Cryptographic problems are the core of most mining algorithms, hashing algorithms, and also of most encryption schemes. This is potentially dangerous. Our current position is: we don’t support this currently, because that would create huge Market impact on a negative part. What we are currently considering is using our technology more for developing better post-quantum encryption schemes and helping researchers with that, but we will not allow any other use case, in terms of reverse engineering hashes, for example, or breaking encryptions through the use of our technology. 

Great answer! I’m sure many people are very curious about that. Would it be convenient for you to tell me how many sales people do you have to solicit potential customers? How big is the entire potential market? From my personal perspective, I think it’s pretty much everyone, definitely governments. Has there been any specific market demand research? 

Some of these of these questions I already answered before. We are in an exponentially growing market. The markets which we are approaching, or addressing, is about to reach more than 1.5 000 billion by 2030, so it’s huge, and the CHR is on average is 38%. This is the most attractive business segments to be in. Key account management, as you’ve seen on the roadmap, is going to start with three regional managers, one by country, or territory basically. We do Europe, Asia Pacific, and then from there we’re going to build it up because we’re always going to have key account management, but, obviously, as soon as we have the marketplace in place there will be some self-service customers too. 

Nice. I’m having a little personal question coming in. How are you hiring these key account managers? Let’s say, if we if we would like to participate more in the community, can we apply anywhere?

Absolutely! Anyone who is interested to join us, help us, and to really be part of a project, which potentially can have significant impact on society as a whole. PM us. We have actually found already quite a number of people which are contributing, and we want to grow it, and feel free to apply. We always need good resources! 

Thank you very much. Let’s get back to the technical aspects. Will there be any stacking pools, farming, in the future? 

Well, short answer is, definitely. I think every proper coin should have these kind of features.

I couldn’t agree more! Any potential partnerships, like.. for example, Flux?

We’re talking with Flux, and as I said before, they’re not the only ones. We have the opinion, we rather embrace competition, collaborate, make something great together, rather than fighting them. So, yes, we believe that it is going to be really interesting going forward.

Can you tell us a little more about what sort of marketing DYNEX have planned? I feel like there is going on an amazing and revolutionary development, but..I know we all noticed, that not so many people yet know about DYNEX compared to something like Tao, which is all over Twitter. How are you going to steer through that?

Well, first we had to prepare the groundwork. Which means, we have a proper technology in place, which is proven. We need to have the exchanges sorted, we have to be present everywhere, and then, of course, marketing is absolutely necessary. I mean there’s going to be a full-fledged marketing and promotion program, as I have explained before, and including everything which is state of the art, obviously. But currently marketing plan is being built by the marketing guys internally and when it’s available, first of all, you’re going to see it everywhere, but we’re also happy to share and there will also be the possibilities to discuss these kinds of things in more specific, or dedicated AMAs going forward great. 

Guys, let’s start tweeting! hahah

Another question we have here. When AMD CPU will be supported? 

Isn’t that my favorite question?! hahah

Before we port the Kuda based implementation for NVIDIA we need to have a solid base. DynexSolve223 is, in our point of view, like good code base for that. 223 is going to be released sometimes next week and then we can start porting OpenCL, which means AMD will be supported because there is a lot of AMD miners out there, which are very eager to join us and we are obviously happy to have them. We think it will take approximately a couple of weeks to port it, and probably a week for testing. Anyone who is interested, who has AMD equipment, and wants to become part of the early testers, please let us know.

We have one question that we all want to know. When are you planning the next exchange listing? 

I mean I answered that before. The plan is the next secondary listing should happen in January.

Do you have any specific date for that? 

We cannot speak about dates.

Okay, then we will be waiting and hoping this is going to be happening soon. And here we have another question still coming in. Do you have any plans for Dynex security hackathon event?

We love that because actually we think we can only grow a network professionally, if you also involve the community. I’m not sure if all of you have seen it, but we do have a bounty program in place. Basically, what we have done is, we divided the bounty by level of severity. From like minor medium to severe, and if you look at the page, you’ll find it on the developer section “BOUNTIES”. We’ve actually granted quite some bounties already. That’s the first approach and then hackathons, is something which is also very, very interesting. We did some in previous projects. It was very successful. So stay tuned! You’re going to see stuff like that happening shortly. 

And on this great note of excitement and waiting for the new developments to come, I think we can wrap up this AMA session. First of all, I would like to thank everyone for listening and joining this channel. I thank you very much for your questions! We will be hosting more AMAs in the future. We will also potentially separate them by topics. Let’s say, to address more the technical aspects because the miners have a lot of questions. So please feel free to post your questions in our Discord Channel. We are reading them all and, obviously, we are paying much attention to them. And thank you so much and I hope to see you next time! Thank you very much for being here.

Thank you for taking your time and participating in this AMA. Thank you for interviewing me. You know I’m available, text me anytime. AMD please PM me and if my response time is a little bit lower over the next few days, it’s because I’m going to Davos and I’m going to be offline for a little bit, but I’m going to check from time to time, obviously because there is some other great stuff coming up. Thanks everyone! It was a pleasure and let’s make DYNEX a really successful company together! 

Thank you so much and safe travels! Thank you very much everyone and hope to see you, to hear you, soon again. I hope you enjoyed this this live session and have a great rest of the day afternoon or morning! Bye, Bye!