Welcome to tech Stuff, a production from I Heart Radio. Hey there, and welcome to tech Stuff. I'm your host, John van Strickland. I'm an executive producer with I Heart Radio and I love all things tech. And Twitter user Janusian w O seven asked if I could do an episode talking about LHR versus non LHR or f HR graphics cards, And that's actually a gateway question to really dive into the proof of work style cryptocurrency mining operations. So today we're gonna talk about what that process actually involves, how that process has affected the graphics card industry, what a s i C is, what an fpg A is, and how companies like Nvidia are trying to respond to the issue so that gamers can act they get their grubby little hands on graphics cards because, in case you're not familiar, uh, graphics cards are sometimes very hard to get hold of, especially to get hold of at their market price. They often end up on secondary markets at ridiculous markups. So we'll also touch on some other elements, like the proposed stopping the grinch bought acts in the United States, But first let's talk about what it takes to mine some cryptocurrency that uses the proof of work approach. Now, first of all, what the heck does mining mean? I mean, cryptocurrency is digital, right, It's just a digital record when you get down to it. So when you really boil it all down, you ultimately get to machine language of zeros and once. How the heck do you mind zeros and ones? Well, since the whole intended point of Bitcoin, which is, you know, kind of the the cryptocurrency that really started at all, at least as far as the current craze is concerned. If you assume that the creators of bitcoin were being sincere they wanted it to perform as a currency, well, that means you have to have some sort of systems in place in order to make this happen. So let's take Bitcoin as our example. Though, just to be clear, the issue with graphics cards doesn't relate to bitcoin. Bitcoin has outgrown graphics cards. When we talk about cryptocurrency, mining and graphics cards, we're more frequently talking about ethereum, not Bitcoin. But we'll get more into that later in this episode. But Bitcoin is the easy one to talk about, right because that's the one that everyone has heard about. A lot of people have heard about ethereum, but not as many as have heard about bitcoin. So in order for bitcoin to work, you need to have a record or ledger of who has bitcoin and who is transferring bitcoin to whom. So without a verifiable record of all the bitcoin that are in circulation, it would be possible for someone to duplicate a bitcoin and to spend it repeatedly. Right, It's just a digital file. You can copy digital files. I'm sure you've done this thousands of times, whether it's to copy a file from your hard drive to go also onto a network drive or maybe onto some form of physical media, or maybe you're just copying a file on your own computer multiple times for some reason. I don't I'm not judging you anyway. Duplicating files is easy, Like it's just a couple of clicks on your mouth, right, So that would mean that you have to have a way to counteract that for digital currency, otherwise you could just print your own money essentially, And they have to be able to prevent people from copying or counterfeit bitcoin, otherwise the whole system would collapse. So one element that helps with this is a distributed ledger of all transactions. To be involved in bitcoin mining, you have to connect to the Bitcoin network, and everyone connected to this network can see a shared ledger that records each and every transaction dating back to the origin of bitcoin itself. Actually, the transactions propagate outward from their point of origin throughout the network. It can take a little bit of time, as much as fourteen seconds. In fact, that is an issue that is a limiting factor for bitcoin in particular. But it's a that's a that's a matter for another podcast because it goes a little beyond what we want to talk about today. So Bitcoin protects itself in part by having this ledger visible to everyone in the system, so that it would be pretty much impossible for someone to to cheat the system because everyone can see the history of transactions. It gets more technical than that, but we'll leave it here. However, you also have to have a way to verify transactions to authorize payments. So let's say you're going to a brick and mortar store. This is just like a an example from everyday life. So you're going to a store physically yourself, and you fill your shopping card up with whatever it is that you're buying, and you're going to use a credit card to pay for it. So when you go up to the cashier and you swipe your credit card or inserted into the machine, whatever it is, the system does a quick check to make certain that you have the available credit in order to make that purchase. So in this case, the point of sale at the cashier would send an electronic request to the appropriate bank or processor for that particular merchant via a phone line or an internet connection. So the request goes from the merchant to a bank. That bank then forwards the details of the credit card to a credit card network like Visa for example, that can then clear the request to send a message on to whatever bank issued the credit card, and it's this bank that authorizes payment if you have the available credit if everything lines up. In other words, so if it's clear that the credit card is is a valid card and that there is UH credit on it. In the case of filling out stuff online, it would check things to make sure that the building address that you filled out matches the one that's in the records, all that kind of stuff. So, assuming everything is cool, the issuing bank since the approval backed down the same chain, so the approval goes to the credit card network. The credit card network then sends that over to the merchant's bank. The merchants bank then sends it to the merchant and says transaction complete. The money has been transferred from the airline of credit to you, and the merchant then prints up a slip that's usually either you sign it or it's just a you know, receipt. UH. This whole process takes very little time typically with the classic system. Right, We've got incredibly fast communication systems, so assuming everything's working it, there's rarely much of a delay. You might be standing there for a couple of seconds while you're waiting for the approval to go through, but it doesn't take long at all. UH. There's also transaction fees that are involved with this. That's important too because it's going to play into cryptocurrency. So with this approach, merchants who accept credit cards also have to pay transaction fees on every sale that involves a credit card. UH. It's usually a small percentage of whatever the overall price is for the items, and merchants typically seek out the cheapest available processing fees, or they might also you know, increase the price of their products to help compensate for that expense, or they'll do both. So credit cards have these centralized authorities, the issuing banks. They ultimately authorized transactions and they verify that money has changed hands. So again, in this case, it would be a line of credit issued to the card holder and then transferred over to the merchant that amount of money. But one of the big selling points of cryptocurrencies like bitcoin is that there is no centralized financial authority at the middle of all this. There is no issuing bank, but you still have to verify transactions otherwise we get back to that problem that someone could duplicate cryptocurrency and crash the whole system. So the way bitcoin does this is it groups transactions together in blocks, and these blocks form a chain, a block chain if you were, and bitcoins blocks currently have a limit of one megabyte in size. That's how big a blockchain block in bitcoin can be. In twenty nineteen, a simple transaction from one Bitcoin wallet to another would take up around around the bytes. Now, if you do some math and you take one megabyte and you divide it by three eight bytes, you end up with around two thousand, seven hundred and sixty transactions. A little less than that, actually, uh, that can fit in one block of the Bitcoin blockchain, so two thousand, seven sixty or so uh per block. By the way, this leads us to some serious limitations of bitcoin when it comes to actually using it as a currency, but again that's the story for another matter, another podcast episode. So, the way the bitcoin system currently works is that a block of transactions are to join the chain every ten minutes. So really the goal here is that every ten minutes you add a block, whether it's full of transactions or not. So even if let's say it's a super slow time and there's not that many transactions, you still need to add a block every ten minutes, because that's just the way the system works. So you have all these computers that connect to the bitcoin network, right these are uh and then you have all these transactions that are waiting for verification. So every time a bitcoin has changed hands, as it were, all of that has to be verified. So the computers that are connected to the bitcoin system, the miners, in other words, when you really break it down, their auditors. They are auditing the list of transactions to verify that they are all valid transactions, that in fact the entity that transferred bitcoin to another entity, that all of that is accurate and valid that in fact there was that bitcoin to transfer, that the wallet it came from is valid, and the wallet it's going to is valid. The computers connected to the bitcoin system are all trying to process this at the same time. They're taking on a role that would normally go to a centralized bank authority, and instead of a merchant paying a transaction fee of the auditors, you've got the bitcoin system itself releasing some bitcoin into circulation, and that that circulation goes straight to whichever auditor is able to verify the transactions first. That's the mining element here, So when you mine bitcoin, what really means is that you were able to validate a block of transactions before anyone else could. Now, interestingly, bitcoin mining will only last as long as there are bitcoin left to be mined. But there is a finite amount of bitcoin. There didn't have to be, you know, you could have created a system where there is no limit, but that's not how bitcoin was designed. So there's actually a cap of twenty one million bitcoin, and once they are all mined, there will be no new bitcoin going into circulation. Interestingly, the number of bitcoin released upon verifying a block of transactions goes down by half every two d thousand blocks, which because a new block has added every ten minutes or so, we can essentially say that around four years, you see the number of bitcoins released per block, mind go get cut in half. So when bitcoin first launched back in two thousand nine, if you were to mind a block of transactions, that would get you fifty bitcoins. So if your computer was the first one to mind it, you got fifty bitcoin. Now, back then, bitcoins were worth a fraction of a dollar, right, fifty bitcoin was nothing back then. Today a single bitcoin is worth more than fifty seven thousand dollars at least as I record this episode, that's where the value is. It's above fifty seven thousand dollars per bitcoin. So imagine that you were able to get fifty bitcoin for mining a block. That means, assuming that the value was the same back then it wasn't, but let's say it was. That would mean you would pull in two point eight five million dollars just for verifying block of transactions, and another block would join every ten minutes. So let's say that you had built out the best mining system to verify transactions, and if bitcoin had been as valuable back then as they are today, then you would soon be a billionaire, except, of course bitcoin weren't worth nearly as much back then. In fact, people would spend thousands of bitcoin just to buy a pizza. It's a very different world. But the number of bitcoin released per block has gone down since two thousand nine. In it was twenty five per block, so went from fifty to twenty five. It reduced again to twelve point five bitcoin per block. Mind today it's at six point to five bitcoin per block. That process will continue until nearly all bitcoin or mind. The fact that you're having and having and having means that eventually you get to a point where it's like, not all bitcoin will actually be mine. There'll be some that will rum unmind, but it's like a tiny, tiny, tiny fraction, right um. But that means that once you get to that point and there are no more bitcoin to mind, then auditors will not be paid out in mind bitcoins because they're they're all gone. Instead, they'll be paid in transaction fees, very similar to the way that merchants have to pay transaction fees on credit card purchases in their establishments. Anyway, that whole system of bitcoin, where we're only looking at transaction fees. That's the good ways off. In the future, it won't be tillt before all the bitcoin or mind. So for now, the promise of mind. Bitcoin provides an incentive to miners to dedicate their computer processing power to verify blocks of transactions. And when I say computer processing power, I'm not talking about the desktop or a laptop. I'm talking about specialized machines that are drawing enormous amounts of power in order to process this as fast as possible. Now, the mining process fulfills some very important functions. It creates a predictable rate of adding more bitcoin to circulation. We know that every ten minutes we're gonna get six point to five more bitcoin, at least until we hit that next milestone, in which case it will be had yet again. And it also brings in auditors to perform the task of verifying transactions, which is necessary for the system to really work alright, So mining ultimately boils down to someone using a computer or these days bloods of specialized computers to comb through a block of transactions and verify them before anyone else does. And then the winning computer system gets some bitcoin for their trouble. But when I say, verifying it's kind of misleading, like that's one way to look at it. But really, what the computers are ultimately doing is trying to come up with a hash value that most closely matches one created by the bitcoin system. So we have to talk about hash values. And yes, this does tie in to those LHR graphics cards. So hash values are really good at establishing the identity of something like say a file. Uh so this gets a little confusing, but just stay with me, all right. So ultimately, you know, a computer file, just like any other computer information is data that you can ultimately reduce down to binary to zeros and ones using a hashing algorithm, which is essentially a set of rules that say this is the process of operations you should perform on any incoming data. Right, That's what an algorithm does, is just algorithms are just instructions. So a hashing algorithm is a specific type of algorithm where a machine takes this binary data these zeros and ones that represent a file, performs some mathematical operations on that data and generates a hash value. Hash value is a fixed length value, and it can you know if it's hexadecimal, it can contain numbers and and other characters as well. Um, so a lot of hashing algorithms convert data into hexadecimal values. I'm not gonna go deep into hexadecimal here. The important bit is that you have a fixed lengths set of values. That means you have a fixed number of characters that will always appear as a hash value no matter what you fed into the UH. The algorithm is going to produce an answer or a result that is that number of characters long. So let's say it's sixty four characters long. That means you feed a big file into this algorithm, it will generate a result that's sixty four characters long as a hash value. You feed a tiny file into the same algorithm, it will also generate a different sixty four character lengths UH hash value. So another words, just looking at the hash values, there's no way for you to tell what the size of the file was, let alone the content of that file, what type of file it was. But it will, which is why it's used on a lot of security measures right because just glancing at it doesn't give you enough information for you to draw any conclusions about what was fed into the algorithm. So again, let's say that we've got an algorithm that puts hash values that are sixty four characters in length. It doesn't matter how you know how large the file is, if it's huge or if it's tiny, you still get a sixty four character length answer, and you will always get that answer based upon the input. So if I have file A and I feed it through the algorithm a hundred times, I will get one hundred identical hash values. But if I feed file B into the algorithm a hundred times, I'll get that one same times, but it will be different from file A. Right like, the hash values are unique to whatever the input is. So hashes are also not reversible. You can't get a hash value and then magically reverse it back to the original file. So if you sent me a hash, I couldn't do anything with it other than really, you know, look at it and compare it against hash values that I have. I can't. I can't reverse it and say, oh, you took a copy of pac Man and you hashed it. I wouldn't know. So what you can do is you can use hash values to verify that a file, for example, that someone has is the exact same file that you have. You both use the same algorithm. They hash their file, you hash your file. They send you the hash value, not the file itself, and you compare it to your hash value. If it's the same, then you say, ah, we both have the same file. That's a way of verifying that someone is who they say they are. For example, like if you ask them to prove that they are who they claim to be, and they send you a hash value that could only come from them because it's unique to something they possess, and you have a record you have like a database that matches this, then you can say, ah, ha I got it. You're in Okay, that's basics for hash values. We're gonna take a quick break because I've been going on a long time here. When we come back, I'll talk a little bit more about hash values before we move on. Okay, I talked before the break about hash values and how you can use them for verification processes where you don't really know the content that was in the original piece. Necessarily you just can compare hash values to each other. Well, hashes are used in lots of other ways. So for example, some anti virus databases will take known malware and they'll create a hash value from the malware, and then anti virus saw where will have a whole database of all these hash values, and the antivirus software will search a suspected infected computer system and hash the various files that are on the computer system, compare it against that database of known malware hash values, and if it finds matches, it says, aha, your computer has been infected by the specific types of malware. All right, but let's let's go back to bitcoin and talk about what hash values mean for that. So, for each block of transactions in a bitcoin blockchain, the bitcoin system itself assigns a sixty four digit hexadecimal hash, and the auditors the miners are trying to generate a hash that is closest to the one provided by the system without going over. And the more miners that joined the system, the more possible solutions there will be to the hash value. So as more auditors join, as more computational processing power is added to the Bitcoin system, the bitcoin system increases the difficulty of the guessing game problem that all the computers are engaged in, and that means you have to have even more computing power if you want to be the one to first mind a block of transactions. Now, computing power is expensive, both to acquire all the pieces for the computing power and to maintain it. Right, you have to pay for specialized hardware because if you don't, then others will. Then you'll be left behind. Right the older stuff that you're using won't be up to task. It would be like if you were trying to enter a street race and you were driving like an old Volvo and everyone else is driving supercars. You would do it doesn't matter how good a driver you are, your your equipment just won't be able to keep pace and uh. And since bitcoin mining is effectively a race to guess a very complicated number as quickly as possible, being left behind means you're stuck with expensive equipment and a reward. So now you've got you know, costs, but you're not bringing money into offset those costs. You would be out of business. So there becomes this escalation among bitcoin miners to get hold of the most advanced machines possible that can work at screamingly fast rates in an effort to mind blocks before anyone else can so. Right now, a block is worth around three fifty six thousand dollars. I got that by multiplying fifty seven thousand dollars per bitcoin times six point two five bitcoins per block, and you get a new block mind every ten minutes. So a day has one thousand, four hundred forty minutes, and that means there are one hundred forty four ten minute blocks per day. So let's let's say hypothetically you create the best killer mining system and you improbably managed to beat out everyone else in the world for every single block in one day. That would mean one day's mining would let you fifty one million dollars at the current value of bitcoin per day, fifty one million dollars. That is a heck of an incentive. Now, this is why you hear about huge bitcoin mining operations that co locate with power plants, because to run those machines, you need a lot of electricity, and electricity costs money. Right Like, if you're if you're paying for your electricity, that bill is an operating expense you have to factor in. And while you could theoretically net around fifty one million dollars a day, in reality, you're you are competing against lots of other big bitcoin mining operations around the world. You're not going to succeed with every single block, So bitcoin mining becomes a balancing act. You know, can you make more money then you have to spend to cover the costs of operation. Can you make a profit? So for a lot of miners, the answer is yes. I mean, like, the value of bitcoin is so high that it's worth going to these extremes to try and go after it because the payoff is better than what the costs are. Right. So, we've talked about how bitcoin mining involves lots of computers all racing to guess the number that will produce a hash that is equal to or less than the hash value that the bitcoin system generates. I'm not going to go into further detail with that because, um this episode is going to be super long already. But but the important part here is that, you know, the more miners that join the system with more computing power makes these problems get even more difficult, which means the bitcoin miners have to feed even more computational power into the system if they want to be competitive. It becomes this vicious cycle that will continue as long as the value of the bitcoin is greater than the cost sunk to chase. After that bitcoin. Right, if you get to a point where the value of the bitcoin is lower than that, then obviously people are going to ease off because otherwise you just lose money. Now this we'll switch gears now and talk about graphics cards. So again, graphics cards are more relevant to ethereum mining than bitcoin mining. Uh. Ethereum one point oh follows a similar approach to bitcoin, and that it's a proof of work system. So that's why I focused on bitcoin first because a lot of what we talked about with bitcoin also relates to ethereum one point oh. Alright, So when it comes to computer processors, there are a lot of different kinds. So your basic central processing unit or CPU is a microchip that has one or more core is dedicated to executing commands operations in other words, on input data. You know, numbers. So a very oversimplified approach might say, like, take whatever data comes in and then add this specific number to that incoming data and produce a result. That would be a very very very simple operation. Right now, your basic CPU has to be flexible. It needs to be able to process information for lots of different kinds of programs. It needs to be able to allow anything from a spreadsheet program, to a media player, to a web browser, to complex computer simulations to run on a machine. These are general purpose microchips. In other words, they need to be able to do pretty much anything. But there's a trade off with this. While a general purpose processor can pretty much do anything, that doesn't mean it can perform any specific task better than a microchip that is only nt to handle that specific task. So, in other words, your basic CPU lacks optimization for particular tasks because the CPU has to be concerned with everything, not just that one thing. Enter the graphics processing unit or GPU. When these emerged a couple of decades ago, the whole intent was to offload calculations relating to computer graphics that would normally go to the CPU and instead redirect those to the GPU, so the CPU would be freed up to work on other stuff, and the GPU would be optimized to do the kind of work needed to produce more sophisticated graphics. So one element that you find in processors is called an a l U or arithmetic logic unit. GPU cards typically have a lot of a l U s, and it's this specific element that becomes really useful when you're trying to perform calculations to guess the right hash value when you're trying to mine a block of cryptocurrency that's on a proof of work style blockchain. The GPUs are great at performing repetitive tasks at very high speed. That's what they were built to do. So Bitcoin miners originally sought out powerful graphics cards to build bitcoin mining rigs. Uh And there are some cryptocurrency mining rigs no longer bitcoin Bitcoin outgrew graphics cards, and again we'll get to that. But like ethereum, mining rigs that are essentially a computer case that holds as many graphics cards as the motherboard can support. And typically these are part of larger networks of similar systems, all with similar buildouts. So you end up with a very high demand for graphics cards to mine ethereum. Moreover, as more miners jump on to try and mine you know, ethereum or bitcoin, it creates a demand for the fastest machines on market. So it's not just good enough to have a ton of graphics cards. Let's say it's ethereum, it would be graphics cards. It's not just good enough to have a lot of graphics cards. You need graphics cards that work at the best rates, at the best efficiencies possible. You're possibly modifying them a bit. If you're new to it, you might over clock your GPUs. That means make them run faster than they were meant to run uh in order to try and process this information faster. But interestingly enough, experienced miners typically won't over clock their their GPUs. In fact, they might underclock the GPU because they're more concerned with efficiency than they are with speed, so they'll still get really fast GPUs, but then they'll reduce the amount of power going to the GPU to make them more efficient. Now, this means that crypto minors have a big incentive to rush out and grab all the best graphics cards that are available, So when a new graphics it hits the market, crypto minors are among the most let's say, enthusiastic customers. Meanwhile, graphics card companies are producing products based on their own manufacturing capacity. In other words, you can't make more cards than what your facility can produce, right, I mean that just makes sense. You can't, like, if you're limited to cards per day, you can't magically produce five hundred cards in a day. So we often see a problem where the demand for a new graphics card is way higher than the supply and the rate of production is such that bitcoin miners will keep on scooping up as many cards as they can until something better comes along, and then they repeat the whole process. Meanwhile, gamers find themselves struggling to track down a new graphics card. This also creates a pretty vicious aftermarket for graphics cards, because there are profiteers who will rush out and buy graphics cards as fast as they possibly can in an effort not to join bitcoin mining. Like they're not necessarily interested in mining bitcoin themselves instaid, what they want to do is sell off the cards they have bought at an incredible markup, like on eBay or whatever. These folks are exacerbating an already tough problem, and they probably don't really care if the person who's buying from them is a gamer or if it's a crypto minor. They're just looking to make a profit. And this brings us to the stopping the Grinch Act proposal in the United States. See One way to get an edge on others is to develop automated scripts or bots that shop for specific items faster than any human could possibly manage. So the idea is that you design a bot to go after something specific, like the latest graphics card from a specific manufacturer, and you buy up all the available stock as fast as you possibly can. Now, maybe you're doing it because you're a bitcoin miner and you just want to dedicate that to your sister. Stim Maybe you're doing it because you're a profiteer and you just plan to sell the cards to whomever wants them on the aftermarket for a huge markup. Either way, you're beating regular human customers to the punch. Well. In twenty nineteen, several US politicians introduced proposed legislation that would make it illegal for people to use those kinds of automated scripts or bots to get the jump on normal retail customers. And they call it the Stopping the Grinch Act, because the Grinch is a character from Dr SEUs who sneaks into a town on Christmas Eve to try and ruin Christmas, which partly involves him stealing all the presents in the town. While the two thousand nineteen version of this legislation stalled out in committees in Congress, but that was back in twenty nineteen. That was before, problems like the pandemic and the semiconductor chip shortage really made things worse now because of supply chain woes and shortages. The fear is that people using these automated scripts will make a tough situation even worse for millions of people who are just trying to do stuff like buy some gifts for their loved ones for the holidays. So, should this legislation pass, the Federal Trade Commission or FTC in the United States will be in charge of enforcing this law, and the hope is that this would create a more even playing field for everyone seeking certain retail items without fear that some automated bot is gonna scoop all of them up before you even have a chance to go to the store virtually, and we'll end up scalping these at a much higher cost down the road on other you know, sites like eBay. Of course, we might just see this legislation dying committee the way it did the first time. Okay, I'm gonna take a break here. Then then we're gonna join in with some more information about crypto mining and graphics cards. But first I'm gonna catch my breath. Okay, let's talk about the issue of ethereum miners going after graphics cards. Now, according to the John Petty Research Group, crypto miners bought up of all GPUs all GPUs in the first quarter of one, so one quarter of all GPUs on the market were claimed before gamers could get a chance at them. Now, that might sound like a bummer, but not too bad. I mean that means three quarters of GPUs we're not claimed by crypto miners like ethereum miners, right, But then you have to remember that crypto miners were targeting the best of the best, the highest end cards, specific graphics cards, which meant that a lot of gamers were left with cards that weren't at the highest of performance levels. And it's not that much fun when you're a gamer and you're trying to really invest in building out a good gaming PC. Then you find out you're gonna have to settle for less and the best because there's just none of the cards you want are available. Also, sometimes miners will choose to sell off used GPUs. This could happen for lots of different reasons. Let's say that the value of a certain cryptocurrency like Ether goes down. Well, if it goes down enough, it could get to a point where the cost of operation is so high that it's no longer profitable to run a large scale mining operation. You know, you'd be paying more money to run things than you would be making in mining. So that's one time when people might start selling off used GPUs. Or it might be that the miner is changing things up, maybe they want to switch to a different GPU, or maybe even the different system altogether, And so occasionally miners will sell off the GPUs they had been using to mine crypto. So the question is should you buy a used GPU from a mining operation, Well, you gotta keep in mind that those GPUs have been run for twenty four hours a day, seven days a week, for however long the operation was relying on them. Now, it's possible that a used GPU will be in good condition and that you'll be able to run it just fine, and that there won't be any appreciable difference in its useful lifespan. However, it's also possible that the mining operation was cramming as many components together as they could in order to conserve space, and if the card received insufficient cooling, because these cards do generate a lot of heat, and heat is bad for electronics, right, Well, not only are these cards generating a lot of heat individually, they're packed in with other cards that are also generating heat. So these these bitcoin mining operations get real hot if they are not sufficiently cooled down while then those cards could be damaged right just from being exposed to heat for so long. So that's something you need to look out for. It's possible that used GPU from a mining operation really doesn't operate at what it should anyway, because just the wear and tear and the heat wore it down. So it really all depends on the mining operation and how it was set up and how it was maintained. Also, you're not likely to see a big mark down in price. Yeah, it's a used GPU and it was used for months, but that doesn't mean it's going to be cheap. Doesn't mean it's even gonna be less than what the market prices for those cards, because they are still in such high demand. I mean, GPUs are are in such high demand that even a used one will end up being valuable enough that people are willing to pay market price or higher than if it were brand new, because again that the supply is so low compared to the demand. Uh, this is dependent upon the actual card and the region you're in and all that kind of stuff. But yeah, it's a harsh reality. Now I've said it a few times in this podcast, but when it comes to graphics cards used in crypto mining, we're typically talking about Ethereum and Ether rather than Bitcoin. Ethereum has some similar at east to bitcoin, but there are also a few major differences. For one, Bitcoin, as I mentioned earlier, has a cap of twenty one million bitcoin. There will never be more than twenty one million bitcoin in circulation. Ethereum does not have a cap. There are actually more than a hundred million Ether coins out there today and there's no cap on supply. Bitcoin's value, as I record this podcast, is more than fifty seven thousand dollars per coin. One Ether is equal to around four thousand, seven hundred dollars. Now four thousand, seven hundred dollars. That's a lot of money, but it's less than one tenth of what you would see with bitcoin, right and you only get to Ether per mind block. Remember right now, with Bitcoin it's six point two five bitcoin per block. Ether you only get to per block. However, you can also mind blocks more frequently in Ether than you can with Bitcoin, because with Bitcoin it's unblocked every ten minutes. With Ether it's one block every fifteen seconds on average. So you do get smaller payouts per block with a bit more than nine grand in value of ether mind per block, but you also have way more blocks mind per day, so it is a different beast from Bitcoin. Graphics cards are still the rage when it comes to mining Ether, but for Bitcoin many miners have moved on to heavier artillery in the form of a S I C S or a SEX If you prefer and FP G A S or FOOTCAS. I guess you don't. I guess you don't actually pronounce the the initialism. Okay, So, and a s I C is an application specific integrated circuit. So remember how I said CPUs are general purpose machines and GPUs are more specialized machines. Well, the A s I C is even more specialized than a GPU. This is a cir it that has been hard coded to do a specific task. It cannot do other stuff because that's just not how the circuit is wired. So an A s I C can be optimized to do something ultra specific and do it wicked fast. A lot of bitcoin miners have spent a huge amount of money developing A s I C machines. They are more expensive than your typical computers are, and they have finally tuned these A s I C machines specifically to generating those hash value guesses in order to mine bitcoin. It's not something that would necessarily be economical for ether mining because again, the scale of the payout is different, so you wouldn't necessarily use these to mine ether. There'd be too expensive to acquire and maintain, you wouldn't get enough of a payout for it to be worthwhile. But for Bitcoin it makes total sense because of the scale of the value. As for f p g A, that stands for fuel programmable gate array, this is another type of integrated circuit. It is very similar to a S I C with one major difference, and that is with a s I C, what you have is what you got right. It's kind of like the old style video game cartridges. You can't write to it, you can't change it. You're limited to whatever was hard coded programmed on the circuit board itself. But an f p G A is at least slightly alterable, slightly programmable so you can make changes to it. Only recently have fpg as started to reach a level of performance that puts them on par with A S I C systems. Before that, they were considered much slower and less efficient than A S I C s, so you didn't see them as much in bitcoin mining operations. That's starting to change, by the way, the bitcoin situation eventually will change dramatically, because either we'll get to a point where the number of bitcoin your that you mind per block is gonna drop to a level where it just doesn't make financial sense to run expensive operations, or we'll get to a point where the miners all become just auditors because there's no more bitcoin to mind. They're only paid in transaction fees, and it wouldn't make sense to run these super huge bitcoin mining operations anyway, and they'll all fade out. So it's hard to say temporary problem, because it's a problem that could potentially last beyond all of our lifetimes, since bitcoin will continue to be mined until but it isn't a problem that lasts forever with bitcoin right. Also, we have to remember that as the amount of computational power reduces in the bitcoin system, the problem. Uh, difficulty will also reduce, so it will become easier to get to that hash value uh as more computers drop out the bitcoin system. So it is one of those things that will correct itself in theory over time. Obviously, if the value you a bitcoin continues to skyrocket, this will persist longer and longer. Right, I mean we know that in a couple of years the number of bitcoin released per block is going to be cut in half yet again, right, Well, if it gets to a point where that number of bitcoin and the value of bitcoin falls below what would mean operational costs, we'll see a drawbag change. But if the bitcoin value goes up, like let's say it's a hundred fifty thou dollars per bitcoin, then people are gonna say, no, it's still totally makes sense to keep to keep pushing as hard as we can because the value is so high. So it's depended upon a lot of variables, all right. That finally sets us up to talk about what the heck and l h R graphics card is. So this was introduced by Nvidia earlier this year, and LHR stands for light hash rate l i t e. Hash rate, and you can probably make some guesses as to what that means. Now, essentially, this technology is intended to check for signs that the card is just being used to mine cryptocurrency by generating hash values. So when the card detects this, the idea is that it cuts that hash rate, that is the speed at which the card can make these kinds of guesses, and these are the kind that relate to proof of work cryptocurrencies like Ethereum, one point oh and Bitcoin. So in other words, the card hamstrings itself if it determines that it's being used for mining cryptocurrency. Now that is a huge disincentive for crypto miners obviously. I mean they want cards that will give them the advantages they need to outperform all the other miners that are going after Ethereum in this case, so they would naturally avoid cards that would essentially slam on the brakes instead. Also, the LHR cards have a higher power draw rate. That means they need more electricity to run, So that means you have you have cards that are not going to process hash values as quickly as other cards are, and they're drawing more power than other types of cards are. That means they're also more expensive to operate if you're running them seven. Technically they're more expensive to operate for gamers as well, but your your average gamer isn't playing twenty four hours a day, seven days a week, so it doesn't accumulate as much. It is a bigger expense, but hopefully not as um prohibitive as it would be for a bitcoin or rather ethereum mining operation. Now you might wonder if this LHR technology would also inhibit gameplay, Like if you hear, oh, this graphics card is designed to put out half it's uh it's output under certain conditions. Does that mean that when I fire up you know, the latest game, uh, that everything's gonna perform more poorly, Like I'm not gonna get the graphics quality and speed that I expect. I'm not going to get the phrase per second that I want. The answer to that is no, because that doesn't relate to generating hash values. As long as whatever you're doing doesn't relate to generating hash values, then the cards not gonna kick in that that break system. In other words, now, the video's approach is not perfect. When the company released the g E Force r t X thirty six, they did so with including the LHR limited. When it started releasing it's it's thirty or three thousand line. Some of the earlier ones came out before in Video had developed LHR, so they don't have an inhibitor on them at all. They can be run at the full speed that's f HR full hash rate. But the thirties sixty came out with an LHR limitter. However, in Video accidentally included in a beta driver for the card away to deactivate the limitter. Essentially, it's like removing a limited on a on a car so that it can exceed its quote unquote top speed. Same sort of thing here, Like you could deactivate that and then run the card as a full hash rate card, and that meant that crypto minors could use the thirty sixty without worrying about the LHR slowing things down. And Video went back to the drawing board and reportedly make sure that newer cards will integrate LHR elements closer to the firmware to make it more difficult to bypass it. That doesn't mean that minors won't try to find ways to crack the limitter and deactivate it. Currently, the r t X thirty seventy t I and the thirty E t I cards still are light hash rate only that's the only version that that in videos cells of those two cards. I can guarantee you there are people who are working to find ways to crack the LHR limitters. Okay, we have a little bit more to talk about, but I've been going on for a long time. Let's take another quick break and we'll be right back. So I think this issue with light hash rates and graphics cards and the the desire for miners to get around it is going to be a bit of a c solve problem in the short term. Um, as long as there is money to be made mining ether through proof of work, then people are going to explore every possible avenue in order to make that money, as long as the amount they're investing is less than what they're going to get as a payout. Again, if we eventually, if this all revolves around money, if it eventually gets more expensive to do than you make out of it, then it doesn't make sense to keep doing it. Eventually you'll bankrupt yourself trying to do it. So if in videos implementation strong enough and hard enough to crack, not not perfect, but hard enough that could be enough to push crypto miners to look at different graphics cards and just avoid in video graphics cards because they have to spend time to crack that LHR protection and time is money. If they take too long to crack LHR before the next round of graphics cards are released, well then it's a waste of effort, right because there are better cards that are already on the market. So that's the ideas. It doesn't have to be perfect, it just has to be hard enough where there's no reason to go that route, you can go some other way. And not all companies are working to discourage crypto mining. Both A m D and Intel are not incorporating any sort of crypto limited on their graphics cards. And you might even say, well, of course, why would they. I mean, the crypto miners represent valuable customers. They demand the best performing cards on the market, so their guarantee, right, You don't have to even market to them. You just have to make a screamingly fast card. So while in videos trying to cater to a specific customer base that of gamers, other companies don't necessarily worry about that. Now that could work in gamer's favor if they happen to like in Video's cards, because if the crypto minors all go to Intel an a m D that frees up in video means that there will be a greater supply of those cards and gamers will actually be able to get their hands on them. If the gamers instead prefer a m D or Intel, well, then they're still competing with the crypto minors. Something else that will likely have a massive impact on this whole situation is that Ethereum is actually headed toward a different method than proof of work. That's how you currently mine ether. Proof of work is the same way that bitcoin works. Right that you have these computer systems that are putting out all this this computational power to race to get to an answer first, Well, Ethereum is going to move away from proof of work and head toward proof of steak s T a k E. So in that strategy, participants in the ether community who want to mine ether will have to put a steak of some of their ether toward the chance to earn more ether from mining. So, in other words, they're dedicating some percentage of the ether they own towards this process. And the more you stake, the more you dedicate to this, the better the odds are that you'll earn more ETHER. This quickly becomes a system that rewards the people who have the most already, or at least who are willing to stake the most already. And there's other problems with that, but we won't get into it. But you know, there's no hash value guessing in this approach, so there's not nearly the same need for computational power. That means that that hold scent of structure that is feeding this frenzy around graphics cards falls away, at least for ethereum, so it will no longer tie to graphics cards, and all those miners who are currently pushing so hard to mind ether will be in possession of very expensive and largely useless hardware. Right, they'll have these computers with screamingly fast GPUs that are expensive to run. But since it's no longer a proof of work system, those systems won't won't do them any good. Those computers won't do them any good. Uh. They could switch to mining a different proof of work cryptocurrency, because there's no shortage of different cryptocurrencies out there. There are hundreds of them. Uh. That and many of them still use proof of work. There's some that use proof of steak, but a lot most of them use proof of work. However, a lot of those currencies are built on top of the Ethereum blockchain, so things are going to change for them, Plaut. A lot of those currencies have a fraction of the value of Ether, so yeah, you could use them to mind these other types of crypto. But if you know, like one Ether is worths and one of these other cryptos is worth point zero to five dollars, well it wouldn't make financial sense to dedicate your screaming fast system two getting loose change, right, You wouldn't be making a profit. Plus, uh, you know, the GPU miners aren't likely to scale up to bitcoin, at least not with GPUs, because they wouldn't be able to compete against the bitcoin miners who are using a S I C systems out there, because GPU just can't match that. So the ethereum miners out there are working on borrowed time at least as the current ethereum process works, and once Ether switches is to proof of steak sometime in two at least that's the plan, we should theoretically see a decline in demand for graphics cards from the crypto community. That should, in theory, give gamers a greater supply and lower the costs for those graphics cards. Uh, you know, there won't be as strong of an aftermarket for them. I think that will be a good thing, And it would also mean that LHR might not be as important unless some other proof of work cryptocurrency gains a ton of popularity, in which case the whole process will start up again. Now, I might do another episode down the line to talk about some of the other limitations of cryptocurrencies, with Bitcoin being the prime example, and I might really focus on things like the number of transactions that a system can handle per given unit of time. Bitcoin has a scaling issue when it comes to processing transactions. The limitations on the size of the blocks in the blockchain and the fact that you only generate a new one every ten minutes means that Bitcoin is limited to processing a fraction of the number of transactions per minute that a centralized company like Visa could handle. Like I think it's like three eight times slower something like that, Like it's it's it's it's just in other words, for Bitcoin to work as a currency, so that people can use it for regular transactions, then it has to be faster at processing those transactions. Otherwise you're waiting forever for a transaction to be verified, and that slows everything down. The entire capitalist system that we use right now is built on speed, and this would be a reversal of that. However, that's kind of a moot point, at least at the moment, because right now bitcoin is really more of a speculative investment and it's less of a currency. So it kind of doesn't matter because we don't see the number of transactions reaching anything close to what VISA handles because people aren't buying and selling or they're not they're not spending bitcoin at that kind of pace. Uh, they're more people are investing in bitcoin. They're buying it, but then they're holding onto it and hoping that that investment grows up, and it might not be for another century or so before we see that change, so we've got some time anyway. My hope is that the combination of LHR technology and the oncoming merge of ethereum one point oh, which is the proof of work version, with Ethereum two point oh, which is the proof of steak version, well, mean, we'll see some relief for gamers out there. I know I have been putting off putting together a gaming rig until that stuff is a little less hectic. The combined problems of profiteers who are buying up cards and selling them at huge markups. Then you've got the crypto minors who are champing at the bit to scoop up cards left, right and center. And then you've got the semiconductor shortage and all the supply chain issues that are already made the expensive activity of building out a gaming rig to become prohibitively expensive, at least for me. Here's hoping for better days and better gaming ahead. All right, that's it that you and I know that that went well beyond what the heck is an LHR gaming or graphics card. But really, I think to appreciate why it's necessary we had to go through all of that and to kind of get a handle on the landscape of crypto, it is possible that will we will see bitcoin miners return to graphics processing cards in the future if the value of bitcoin drops to a point where it would be prohibitively expensive to run as a systems or a s I C. Systems, and you would want to downgrade because otherwise you'd be losing money by I mean, that's a Again, that's a delicate balancing act that will be dependent upon the value of bitcoin and what other miners are doing. So we'll have to wait and see how that all shakes out over the next few years. Um, I'm still not like you could say I have a bias against crypto, and I think that would be totally fair for you to say that I am not a huge crypto fan. I feel like it's it's not a Ponzi scheme, at least not in the purest sense of the word. But I do feel like it's yet another system that consolidates an enormous amount of wealth among a very tiny group of stakeholders at the expense of everyone else, and that it doesn't really benefit society in any way apart from like the small number of stakeholders who are making crazy bank on the whole thing. Ah, but hey, that's my opinion, and and I fully admit there are plenty of evangelists out there who love crypto who would disagree vehemently with me, and I'm not saying they're wrong, And then I'm right it's just kind of my own perspective. If you have suggestions for future episodes of tech Stuff, kind of like the one we just did today, reach out to me on Twitter. The handle for the show is text Stuff H. S W and I'll talk to you again really see. Text Stuff is an I Heart Radio production. For more podcasts from my Heart Radio, visit the i Heart Radio app, Apple Podcasts, or wherever you listen to your favorite shows.