We are honored to have Henry Quan join us today. He is the CEO of ePIC Blockchain, a North American based Crypto-mining ASIC design and rig supplier. Henry brings 30 years of high tech and semiconductor experience to the blockchain community.
He was an executive at ATI Technologies for 18 years and helped grow the company to over $2 billion in sales annually and become the #1 GPU manufacturer. ATI was acquired by AMD for $5B in 2006. Henry held various executive roles from product marketing to strategic planning to Corporate Development.
Ironically, Henry and his team, all formerly from AMD, have been lured by away from their GPU roots to “Evil ASIC Empire.” ePIC’s mantra is “ASIC resistance is futile!”
https://www.facebook.com/groups/crypto.mining.tools/permalink/603748550539117/ https://www.facebook.com/cryptominingtoolspodcast/videos/1257172184614850/ https://www.facebook.com/scott.offord.milwaukee.seo/videos/10163721069990024/
Hey everybody. We are alive today. I am your host, Ethan circa, and I have my cohost over here. Sucky is a partner with our company and one of our esteem team members, he, he is indispensable and we wouldn’t be able to get things done without him. Sucky, please introduce our guests today.
Yes. today we have Henry with us and he is from Epic. And he will basically, he, Henry has a very, very long history in the Harbor manufacturing industry. He actually from 19 years ago, he was with ATI. I’m sure a lot of you gamers know ATI and yeah, I’d like memory to just take it away and please introduce yourself.
Yeah. We’re very excited to have you Henry you know, thanks so much for being part of this podcast. I’ve been excited to have you on for quite some time. Tell us about how, how did you get here to where you are in, in, in life today? Yeah. Tell us the history about ATI and, and bringing us up to speed. Sure.
So, you know, I love hardware. I’ve been in hardware for 20 years before this, and, you know, at ATI, we really helped shape the us, the PC business. You know, we’re doing really cool things, such as defining the gaming API, which became the basis for X-Box in know with the PCI local boss. We help Intel develop video. And so all throughout, we were very instrumental in terms of shaping the world of the PC business. Even windows accelerators, you something simple, you think windows runs today, but you know, back in the day when we had dinosaurs and we had to carry our laptops, you know, on our back,
You know, these things were very we’re different windows. We dragged across the screen, they used to jump. So, you know, we helped pioneer the concept of windows accelerators too. And I’m really excited about blockchain because we think that the innovation that has been happening in blockchain is nowhere near what it should be. And, you know, it’s true. I don’t want to say professional firms, but through people that really understand chip technology that can come in and help revolutionize the shape of the business, we’re not in it just for cryptocurrency. Our end goal was to provide what we call a BPU, a blockchain processing unit to replace Oh, wow. Use so that when you can fully run these transactions on crypto, you can run it like a visa in a hundred thousand transactions per second, instead of maybe a, you know, a hundred transactions per second, that we’re getting today.
That is absolutely fascinating and amazing. Can you explain to our audience, what would be the hardware difference? Would you be using some sort of application specific integrated circuit instead of a generic CPU or, yeah. Okay.
Yeah. So basically the, you know, the key difference is that in crypto today, we have mostly hashing. We have hash and chips. So you build one circuit, you replicate it, you optimize it for power and frequency. You know, those are what we call the hash Frisch chips or your traditional you know, cryptocurrency chips. Then you have this next, we have the stuff above it, which is the, you know memory intensive chips. So there aren’t a lot of memory intensive chips today. You typically look at these like an echo hash for instance, or the good Brin. And, you know, the GPU are all in this memory intensive category. There’s a lot of innovations there. And you know, crypto only users are part of a circuit. So when we look at a full BPU, what you get is you get something just more general purpose as adaptable to more algorithms, how to integrate it. You see this trend already in AI chips, where memory is going into the ASIC, you’ll see the same thing happen in blockchain processors as well. That is really fascinating.
And, you know, did you expect that those would be as efficient as you know, that technology would be more efficient than the current technology that we have right now?
Oh, definitely. I mean, basically today you’re talking about general purpose applications. So even to do blockchain today on CPU and GPU and, you know, as good as they are, they’re not designed to, for super fast transactions, you saw Intel put up their SGX extensions. I mean, that’s going to get you some improvement, but nowhere near what you need. So I’m envisioning a room of racks of blockchain processors, which are, you know, a hundred times more efficient than a GPU CPU combination. Wow. Wow. And are you talking about efficiency and just speed or power or both? Both. Both. Wow. That is really that’s, that’s really cool. Do you, do you have anything to show us in regards to that? Well, I don’t have anything to show you in that respect, but I can show you, I can show you what, basically this is a box.
This is half a wafer chips. So a chip, basically a weight for is a 12 inch disc and there’s little squares on it. You cut up the squares and become the chip. So, you know, we’ve got four of these racks in here and I don’t know if you can see. Oh, wow. Yeah, yeah. So each one of those little squares is a processor. That’s correct. That’s so each one of these is the, is a chip that we use on our recently announced C 200 rig. That is really cool. I, I’m always fascinated and loved to see, you know, technology being developed and manufactured and created what it looks like on a board. So, sorry. Cause that one’s hidden in the carrier, so you can see each one of the are here.
Oh wow. And, and is that a hash board? That’s correct. For hashing. Wow. And I can see that there’s very few chips. Yeah. Normally there’s many, many, many chips on there, so, and that’s because they, these chips are so much more efficient. That’s why you need fewer perhaps boards.
That’s correct. You know, we designed, we designed the chips to a certain power envelope. You know, we’re doing the Blake to the algorithm and it’s not, as, it’s not as power intensive as the, as the Shelton 56 algorithms. Are you, are you planning on moving to any other algorithms? So we have we’ve, this is the first of our three three rigs that we have planned. So we’ll have a rig. I can’t talk about them now. Man, I can tell you lots of backs, channel stories about how this industry’s a little bit nefarious in terms of competitive position. I’m sure you’ve run into them as well. Absolutely. You pre announced the product and your competitors then announced a product has better specs, but you have no intention of shipping and they ended up calling your sales. So, you know, that’s why we keep our products close to our close to our chest. So I’ve got, I’ve got two more rigs coming and you know, we’ll go into higher and higher into the cryptocurrency stack as far as pointing copy popularity. Excellent.
That’s yeah. T tell us a bit more about the the woes and the pitfalls of, of the industry, because I totally understand what you’re saying and exactly what you’ve said. I’ve seen happen. I’ve seen one manufacturer come out with a spec miner and then suddenly another manufacturer claims that they are, they’ve got a 2.0 version of it and it never comes to fruition and it does, it does burn the sales of the other minor. So tell us a little bit more about, you know, these kinds of stories that you’ve had. Can you extend the podcasts, but I know the three hours. No, unfortunately not, but we can’t have you on another time. Sure.
We can do that as well. So, you know, there’s lots of pitfalls I can just rambling off the top of my head is you know, I think the, the whole idea that chips are easy. You seen many manufacturers in the past say I can do a chip. So they contract the hardware company, they contract to do its chips. They contract a bore company and they contract the firmware company and then you try and put these things together. And in that belief, it never meets spec because you know, they either scrimped on budget or they don’t set the spec. Right. Or they don’t have even the right people to evaluate how well the system is going to work. You know, when we’re at AMD or, you know, here, we spent so much time up front saying we’re going to develop a spec. Here’s how it’s going to work.
And we, you know, we do the simulations and we have everything in place before we start the project. Otherwise you’re getting this, you know, you’re getting this kind of Lego. It’s like a blind man building a legal, like I said, something’s going to come up, but you have no idea what it is. And so that’s been, that’s been one of the biggest challenges in terms of, you know, and a lot of this is because we have software people defining hardware standards. I don’t know. I don’t profess to know how to build a house. I can tell you generally how it’s done, but you would never contract me to build a house. So why would you contract a software person to help you to help you build a, you know, a rig look at the people that are building it or making the specifications. They don’t have much semiconductor experience.
I spend all my time on forums trying to phew, these claims that are not achievable. So that’s the first bit fall, you know? And so, you know, a company like Epic coming in would be a professional company. If you will, with solid semiconductor experience, we’ve built 10 generations GPU. And you know, we built six genders on game console. So we’re the guys that, you know, you know, you know, we know how to build a high quality of hardware. So, you know, that’s kind of the back end. The finance side is that, you know, because you’ve bought this kind of, you know, inadequate specifications and then you have all the marketing FID is you now have this product or competing with paper tigers. As I call it, people announce a spec and they won’t achieve the spec where they have no intentions to release the product. You know, Intel, Silicon, you wouldn’t expect to do this, but they did the product. They made the same thing with green. You know, we did a, we did a single chip with integrated memory for grin and it’ll Silicon announced their product compete with us, but didn’t really add anything. So examples of how the industry is works.
So Henry, it seems to me that you make it sound like everything is happening all at once. That there, isn’t kind of a linear progression that it’s just software is trying to create software where hardware is trying to create hardware while marketing is trying to, you know, generate interest in sales. Is it really happening like that? Or, or is there kind of more of a linear step by step process?
Well, you can look at the, you can look at the development flow as somewhat our thought now, right? So marketing’s going to look at it and say, here’s the, you know, here’s the market for, here’s kind of the products we need to do engineering. We’ll look at it and say, okay, you know, we know this algorithm can do this. And you know, we think we can get this type of efficiency. We’re going to intersect it with a 16 nanometer process or a seven nanometer process. And we’re condensed after competitor’s parts, we’re going to study the algorithm, see how they can be engineered. And you know, it’s going to be a combination of all those ideas coming together that will make a great rig.
Where, where does the inception come from? Does inception come from the marketing side, listening to what the market wants or does it come from the engineering saying
Here is what’s possible. And this is our next level of improvement that we can move towards. It’s going to come mostly from engineering. Marketing will help determine which coins are popular and you know, and kind of the market dynamics, but really it’s a technical issue, Mark. Okay. Take it so far. So, you know, we can say, we can say, yeah, let’s do it. But engineer might come back and say, you know, this doesn’t make any sense and I’m sure you’ve heard many of those conversations more. If you got paid a dollar for every one of those conversations, you wouldn’t need to work. Am I right? Exactly. There’s a classic opportunity. You should take a look at, it says, it shows, it shows a customer, a kid’s wanting a swing. And it shows me wearing a belt, which is this fancy schmancy swing. And it shows what marketing wants, which is, you know, this thing in the glass house. So, you know, somewhere along the line between this triangles, right. Is the right answer. And now, and yeah, go ahead. So keep it up
Now, Henry, I know I was going to bring up you know, you coined the term async resistance is futile. I kind of want to talk about that a little bit because you know, we’ve had some interest in that in that phrase I wanted to know about the origins and also I wanted to know about your your thoughts on async resistance overall and what you think about you know smaller blockchains that can be possibly, you know, chained off using a thick.
Yep. So basically the way we look at it is, you know, nothing is, there’s nothing that’s, tamper-proof, if you’re gonna, basically, if you’re going to, someone puts their mind to it, they’re gonna figure out how to solve it. Right. So the whole idea on AC resistance is that you, you know, you basically tie an algorithm to memory and, you know, because memory is big and expensive you can basically use the big memory models combined with, with a changing hashing functionality to make it ACP resistant. And you know, a theorem is a good example of an ACO versus the coin. You know, what’s interesting is that grin it’s those basic friendly, and they inadvertently became AC resistant because their algorithms still damn tough to implement. Now grin uses the nimble Wimble algorithm, right? They use Momo for security, but they used to cycle for half. Right. it’s scraps, Ashley’s comes up with these names. I love these names like, wow, I’m actually the engineers talk to their children and say, what should I call this? An interesting backstory, which is the founders or grin loved Harry Potter. So, you know, everything in there is symbols, Harry Potter, the founder agno is, you know, has the, was one of the characters in Harry Potter
And I’ve been a minor of both grin and beam. And you know, I, I like both. I have, I think that both has their advantage has its advantages. So I was wondering so obviously this, you know, your minor would work on, on, on beam as well, right?
No. So you know, that one doesn’t use, the beam uses a what they call, I think Ecqua beam. So it’s a variant. So all the currencies try and make their, make their argument slightly different. So, you know, if you look at, we look at Raven coin, Raven coin adopted Prague pal, but they have their own version of Prague, which is called the path. So, you know, in order for us to make an ASIC resist, in order for us to address an ASIC resistant coin is we have to build the algorithms into the ACEs, right? So there could be nuances, but you know, one little change means that we have to change the AC. So we try and go all this variability in to support the basic resistant clones.
So, I mean, where do you see the future of these coins? I mean, obviously there, there’s a lot of different camps. There’s a lot of very smart people putting in effort and energy to make these successful. How do you think they will compare or fare in, in the next five years compared to Bitcoin?
Well, are you going to censor me for saying this? Okay. So I guess, first of all, there’s way too many coins, right? There’s, you know, a thousand, I think if I look at the periodic table, there’s over a thousand, over a thousand easily, right? When all of them, you know, really only 20 of them have much any practical rag. So the market’s somewhat saturated. I think you’re going to see the stable coins survive because it is always a market for kind of safe type instruments. You’re going to see the coins were big adoption survive. Although I think mining would change dramatically on this. You see a new application of staking, which is just another form of holding a coin. And then I think the ruins I’m really excited about are the application coins. We love CYA for the reason that it really promotes the user.
You know, it’s got a practical application behind it, which is, you know, a blockchain storage. We can see that vehicle easily, you know, easily with the right support vehicle overtaking Amazon it’s, you know, it’s version three, it’s crowdsourced, you know, and it’s got superior technology. I think Saya scream is very interesting from that standpoint that you can do streaming media on it. It’s going to cost like a fifth of what cost to host on Amazon. So all the, all the amateur producers on YouTube and now hold their petabytes of storage on CYA and inside has QoS built in which, you know, isn’t available to the small companies.
I I actually
Mine site dual mining on my GPU is you know, that was mining at 2017 and you’re not seeing them grow into into a really nice you know, option for a decentralized storage platform. Yeah. And I think to come because you know, their, their, their pace of innovation is really stepped up.
Yeah. I, I’ve always been very interested in them as well, but Sikkim if you could, let’s take a moment and let’s thank our sponsor for today’s show.
Yeah, definitely. Well this episode guys has been sponsored by EcoChain an up and coming player in the cryptocurrency mining industry. EcoChain is focused on powering their expanding set of data center assets with a 100% green energy. EcoChain is a new venture wholly owned by a public traded company, Mechanical Technologies Incorporated (OTC-MKTY). They want to secure the blockchain with green and renewable energy that’s good for the planet. Good for the crypto ecosystem ecosystem. And they’re currently looking to acquire additional mining operations aligned with their strategic model. So if you have a mine or you would like to sell equipment, you think we might be a good fit email them and the email address [email protected]ochainmining.com.
And I’m going to go ahead and copy this and paste this into the chat
Yeah. [email protected]
Yeah. Thanks a lot. Thanks. Lots of Q and a, yeah, just likes to keep said you know, they’re innovators in the industry, so, so give them a shot.
All right. Back to, to Henry. Let’s see, where were we? We were talking about algorithms and we were talking about Isaac resistance. And you know, also it’s important to note that the manufacturing is going through a lot of troubles right now because of the pandemic. Yeah, but Scott offer does have a question. So let’s, let’s, let’s answer that question real quick for Scott. Says,
Yeah. Why is it important basically? Is this question,
Well, this inherent belief that, you know, Asics are bad you know, that we see the centralization in China, we see evil practice. We see practices by evil manufacturers, like bit Maine. We’re doing South mining where you keep the product, you keep the best product for yourself. You use them, you know, you gather all the money and when you release your new generation of product, when you have another product coming out, then you release your old products. So a lot of the blockchains want us want to get rid of this domination by six in place, you know, place the power and decentralized the smaller. So, you know, that’s the theory kind of basic resistance. Most basic resistance technology has been designed to fold a number one in GPU. They’re supposed to be widely available, which is a total fallacy because we know that GPS are highly centralized in the same areas that are, you know, also doing mining in China. And I’ll have you, I’ll tell you a deep, dark secret that no one knows. Yes, please do. Okay. You heard it here first, right? Okay. So a lot of the GPU board manufacturers are doing self mining. So why do you think, why do you think there’s such a shortage of GPS in 2018? Why should someone sell a gift for you, a you for $300 when they could keep it for themselves and get an ROI in four months, that’s where it’s happening in a theorem.
Wow. I never thought of it that way. You know, we, we always thought of you know, the, the kind of PC world still existing outside of this space. But what you’re saying is the manufacturers we’re PR we’re basically pre mining, getting an ROI on their cards and then selling their cards to customers is brand new, getting another another income from it as well.
Yeah. Or keeping it too long and having to write it off. Now we see the same thing happening in reverse. Now that GPS is all written down. These are obviously all free and they’re kind of mining in the background. You kind of tie that to your operations as a public company and you end up self mining. So the second reason, the second piece is resistance is you keep changing your algorithm. So in that way, you change your algorithm every six months. And it’s hard for ACP to follow cause Asics are fixed function. They can not change unless you make them somewhat programmable,
And are they programmable? Or you can use that. The chips that we’re working on are not programmable at this point, they are fixed function. Now fixed function is a, this, you know, you typically build a specialized cores and then you can make them run fast. So, you know, somewhere between a GPU, your programmable, blockchain prosecutors, somewhere between a fixed function and a GPU in terms of performance and functionality.
We’ve got Rico Liberty report here saying centralization is being broken up with the last havening and big players moving into Quebec, Georgia, Texas, Nebraska, Argentina, Venezuela, Pakistan, Iran, and other places. What are your thoughts on this?
I would agree with that point in general, but I don’t think I hit the whole thing. Access to technology is the key, right? So fundamentally, we’re going to see a bunch of things happening and in five nanometre the, you know, this is going to go way beyond the scope of this discussion, but, you know, the, the availability of new technology is very limited. From the standpoint that we see Xbox ramping, we see Nvidia seven nanometer GPS grabbing, we see apples, a new process of wrapping. So those are going to swell up most of the, the supply for new technologies. And we’re going to see, we’re going to see the market in Bitcoin stabilize. So that you’ll be much like the S nine when it came out that because seven nanny was not available we saw the market stagnate at S nine levels for three and a half years. You’re gonna see, you’re gonna see a little bit of that happening in the new Bitcoin product. So that’s going to lead to this decentralization you’re talking about, but the moment the five nanny meta parts come out, you’re going to see this big rush back to centralization that, you know, you’re going to use the new hardware to, to obsolete the old hardware.
Now, what, what do you know about the five nanometer process? Because everything that I’m hearing is there’s, they’re still struggling with reliable manufacturing with that. You know, we can only, we can only get the pathways so close together before we start having artifacts like quantum tunneling and things like that. So I mean, how, how much progress is being made on that. And another thing too, is a lot of people don’t understand is people think that if you have the nanometre, you, you double the technology, but that’s not actually true. Can you dispel that myth to everybody? Yeah,
I think the way she plan on this is that you’re going to get third. Traditionally, you’re going to get 30% improvement with each generation, and then you can decide how that split a 30 is improved, right? You can either divert it to a 30% more performance or 30% less power. And then you can optimize with little tweaks, like custom, you know, basically a custom development to try and, you know, improve your power efficiency. So, you know, those, those are the knobs that you turn, and then you can also change the die size of whatnot. But to get back to your question about five nanometers, what’s going to happen. I think you’re going to see TSMC run away with the technology. And, you know, you already see that if you turn for the readers that are interested, I would urge you to refer to Wikipedia and type in five nanometers that’s comparison with Samsung and TSMC in terms of transistor density and memory density. You’ll see that TSMC is depending on which note, which variant a five you compare is between 30 and 50% a denser. That means you can pack more information that the sevens, that’s why you see that, you know, Apple, all the high performance guys in HPC, which crypto is in the category of evolve, gone towards TSMC. Samsung is going to carve this niche in low power, but, you know, crypto doesn’t want low part. We want the highest hash rate, right?
No, I mean, do you think there’s, there’s ever going to come to a point where the density is going to become so dense that like, like right now we’re, we’re struggling to get enough power to the miners. Like you, the C 19 receptacles are rated at 16 amps and, you know, farms are having, they can’t use 208 volts anymore. They have to use 240 plus to get the power of just today’s minor. Now, if we, you know, double the, the, the power demand, I just w how, how do we overcome that issue? Are we going to have minors with two power supplies? You know, doing two hash boards each, or,
Yeah, I think you’ll see, I think you’ll see a change in the form factor. So you’re going to see a lot of this stuff was designed for the whole minor, the toaster, the toaster boxes, asinine size, aren’t the most, you know, efficient way to package a minor. You know, I would, if I were doing it for industrial minor, I would do blades, right. And we can do much better in terms of air flow and form factor in density to address your question about powers. We’re always going to find to improve our lose power, but, you know, the AC resistant coins are probably the best at reducing power cause the memory and the memory aspect of it uses less part than the hashing functions.
That makes a lot of sense. Nick Burley here, which, hi. Hi, Nick. Good to see you here. He’s saying, what do you think the timeline on five nanometre will be for us
Minor manufacturers? Do you, do you want the PR answer or do you want the practical end?
I want the practical answer given the current state of affairs, because we all know that a big manufacturer are having a little fight on the playground and everybody else is running around. They don’t know what to do. They don’t know if they need to call the teacher or the principal. And these two, these two guys are just duking it out. So tell us what you really think.
Okay. I really think that five nanny meter technologies will not be available commercially on the market in Bitcoin until end of 2021. Okay. That’s about, that’s about what I sense too, right? I think we’ll be lucky to hit those dates. So the, the reason is that is that Bitcoin lined on a custom technology and custom technology requires a lot of hand placement. You do those little tweaks, you try circuits, it works, or it doesn’t work. You look at what you Han did in October. I think September, 2018 stood on stage, hold this ship up and said, I will have this in production at the end of the year, where you chips later on the [inaudible] they barely got into production. And you know, those SMEs are failing to feel like, Oh yes, five is denser. There’s no, you know, the problem of custom technology. We have no learning experience. So what you do for seven doesn’t work for five. So you’re going to Excel. You’re going to waste. Basically. We think you’re going to waste 50, $60 million burning mass turning prototypes until you get five minutes meter product running. Okay.
Really? So it’s not just a simply, it’s just scaling an ad, adding more transistors. You, you have to kind of just rethink
Everything. Yes. Potentially
Exponentially more difficult as you get to a five
That’s right? So every, every generation is going to be tougher and tougher to do this gets back to your transistor densities. And the other thing to keep in mind as is the business dynamics of it. So, you know, one, you have no experience. You don’t know how these chips are characterized, so you can design your circuit, but you don’t know how practically it’s going to perform. Cause TMC have the option with data. What they will have is data from their ramp customers. It was just going to be Apple and a lot of what pies, the Apple doesn’t apply to, you know, crypto the crypto mining space. Right. And then lastly, the reason I support that ramp is they’re going to give all the supply to the high volume customers. Don’t forget that bit main burn TSMC big time in the 18. Oh yeah. Did. Yeah. That’s first time that TSMC announced a financial shortfall.
Another thing too that I’ve heard in the industry is that Bitmain is only getting about a 30% return per wafer. So only about 30% of the chips on wafer are acceptable enough to be used. Now we probably know they’re using more. They’re there they’re sprinkling in some bad ones with the good ones, but that’s a huge factor too, because you’re only getting a yield of 30%, but you still have to pay for a hundred percent of the wafer. And one of the groups that we’re working with, they have, what, what makes them unique is they have an 85 plus percent yield per wafer. So do you think that has a huge factor in, in what lies ahead for Bitmain as well? Like
No. No. It just, you know, I don’t know systemically what their problem is other than, you know, I know they got some political problems right now, which is causing it was causing a war between Beijing and Hong Kong grad. So, you know, they can’t, they see a bit men can’t ship. I think that the S 17 is a byproduct of, you know, having chips that don’t make spec. So this is very similar to what you see in the GPU industry of CPU industry. We short you start chips into hi, Ben and Logan in becomes the S T 17 and the high benches. Yes. 17. So you got to run more tests on them. A yield is typically based on defect density. So, you know, you’ve got errors on the, you know, on your wafer and the board dents, basically the bigger chip, the more errors you’re going to get. So there’s more likelihood of getting a bad chip on a yield standpoint, but the Bitmain chips are small. So I don’t think that’s the issue. You know, I think their issues are kinda more, more kind of bending related.
Okay. Yeah. That’s, that’s a very interesting perspective.
Henry, I wanted to ask you as far as, you know the challenges that you’re facing far as, you know, getting
Your miners out there doing this Corona, show us your miner real quick, Kendra. We didn’t get a chance to see it. Ah, how many of you watched MacGyver? Yeah. So we’ve had to do everything to become a MacGyver. You know, we, we we’re supposed to have chassies for our rig and, you know, low to behold, they don’t appear, right. Because the can’t get enough aluminum or the person making the, the dye stamp isn’t there or so wow. Cause we had to start testing actually, that, that looks pretty cool. I wouldn’t mind having one of those. Yeah.
Yeah. We, we, we are, we got the heat sink spec wrong. So when we did the thermal testing, they were off. So we bought these, you know, we bought these from old, you know, from old RF equipment and then one of our guys bought a CNC machine on, you know, on just machine them out there. Right. So here’s the, Oh, sorry. Here’s the old, I got it wrong. Here’s the old, here’s the ones that we machined. Okay. And here’s the production one. Oh yeah. Much better, much better. But you know, these are the things we had to do. Like, you know, we, you, I can tell you stories about COVID bring up recurring boards between each of those houses. We rent it extra Cylus, scopes and test equipment to put in this home. So we could be doing concurrent, bring up. That is, yeah. I mean, this, this will be a story for the ages. And I think we’re so lucky to have had you on today, Henry, and to talk about these things, Henry, can you tell everyone what’s a good way for them to get ahold of you to reach out to you? Sure you can reach me on telegram at Henry underscore Epic. Okay.
Underscore Epic. And is there an email or is that just your, your preferred way? That’s my preferred way. Okay. Excellent. Okay. That’s right. Have the conversations. And then if, you know, if you want it to look at our side coin miner, go to our website under S I think it’s Epicblockchain.io. I see 200. Right. And you’ll get, you know, we put some technical information up there, so you can see some of the differences that we went to in terms of engineering, a minor for quality. And those are some things we’re going to keep doing to make sure that, you know, our customers have the most reliable minors and keep watching us for new minors. Like, you’ll see an announcement from ups and Q1 and, you know, Scott will probably have a lot of these two that’s right. Yeah. Scott, Scott and I are really looking forward to working with and, and your future launches. Yeah. I love Scott’s comments on telegram and he’s, he’s such a, he’s such an interesting character. Alright, excellent. Thank you very much. Thank you very much for having me here. Work with the guys again and I’ll see you. Right. Thanks guys. Bye.