The August 2022 issue of IEEE Spectrum is here!

Close bar

Bitcoin’s Computing Crisis

The computing power of the network that runs Bitcoin doubled in October, pushing out all but the most dedicated (and richest) miners. Could an alternative currency, Litecoin, be the solution?

8 min read
Bitcoin’s Computing Crisis
Photo: KnCMiner

There are many benchmarks you can use to measure the growth of Bitcoin. All of them—the price listed by online exchanges, the number of new merchants accepting the cryptocurrency for goods and services, the transaction volume across the Bitcoin network—suggest that Bitcoin is steadily gaining in popularity. But the most impressive metric by far is the astronomical increase in the processing power of the network of computers involved in running the transactions and creating new bitcoins. This month, it exploded, doubling in just a few weeks the amount of power it had previously taken more than four years to accumulate.

The change has occurred as enriched bitcoin miners have reinvested their profits into new, sophisticated hardware, a trend that’s likely to continue until companies manufacturing this equipment bump up against the state of the art in computer-chip technology. In the meantime, smaller operations, the little guys who once had a decent chance of earning some bitcoins on their laptop PCs, are being edged out by the competition, leaving the stability and security of Bitcoin in the hands of fewer people and threatening the reputation of a currency that was designed to distribute power among the masses. For those who want to opt out of this arms race, or who can’t keep  pace with the upgrades in Bitcoin mining, an alternative already exists: another a cryptocurrency called Litecoin, which was specifically designed to fix the problems Bitcoin is now facing.

But to understand why Litecoin might be a solution, you have to understand why Bitcoin is entering strange territory. The speed of the Bitcoin network is measured in hashes, which are the fundamental calculations processed by miners running the Bitcoin software. The faster they are able to churn them out, the more likely they are to create a new chunk of bitcoins. From 2009, the year Bitcoin was created, until early this year, the cumulative hash rate of all the computers hooked into the Bitcoin network worldwide grew only modestly, never rising above 50 000 gigahashes per second. But when change arrived, it came suddenly. By October, the rate was 1.9 petahashes per second (nearly 2 quadrillion). And in less than a month, that rate has doubled again.

The increase came as mining rigs based on application-specific integrated circuits, or ASICs, hit the market. The ASICs in these computers are specifically hardwired to compute the Bitcoin SHA-256 hash. Though it came late to the game, Avalon was the first company to get an ASIC miner to the public, shipping 300 units by the end of February. The rigs each added about 60 gigahashes per second to the Bitcoin network, a boost that was immediately noticeable. But that was only the beginning. Butterfly Labs finally got its smaller 5-gigahash-per-second rig out to disgruntled customers who had been waiting over a year for shipment, and a Swedish company called KnCMiner sent out a considerably more powerful, 550-gigahash-per-second rig in October.

10NWBitcoin Hashrate f2

10NWBitcoin Difficulty f3 Bitcoin computing skyrockets: The hashrate is a measure of the computing power of the peer-to-peer network of computers that handles Bitcoin transactions and creates new bitcoins. Computers with purpose-built chips shot the hashrate past 1 quadrillion hashes per second earlier this year. The Bitcoin network compensates by increasing the average number of hashes a miner must compute to in order to make bitcoins. But this has the effect of driving away the general purpose computers on the network.Charts: BitInfoCharts

It should be noted that the rigs rarely operate at exactly the speeds these companies claim, with some running faster than promised and some running slower. But the remarkable effect on the speed of the overall network is evidence enough that they are getting the job done.

And rig manufacturers have yet to reach the state of the art in chip technology. Typically, with each new generation of chip-manufacturing technology, the number of transistors that can be squeezed onto a given area of silicon doubles. Each generation is named for a measurement—such as “65 nanometer”—that once referred to the smallest feature on the chip, but that number hasn’t really meant anything concrete for years. Intel now leads the field, having demonstrated a chip made with a 14-nm process, but the foundries that build low-volume ASICs are usually not at the cutting edge.

The first mining rigs that Avalon sent out earlier this year were made with a 110-nm process, far behind the industry standard. The best available is the KnCMiner, which uses a 28-nm ASIC called the Jupiter. Progress has been fast and furious up until this point, but a slight tapering off past the 28-nm mark can be expected.

Big players such as GlobalFoundries, Intel, and Taiwan Semiconductor Manufacturing Co. “are the ones doing the research into lower-nanometer processes, which is very expensive research,” says David Perry, a bitcoin miner in Las Vegas. “As such, they enjoy a period of exclusivity before they release those processes to the rest of us,” he adds. Perry reviews new rig technology on his blog, Coding in My Sleep.

The profitability of a single mining computer is measured by the relative power it contributes to the entire network. As these new rigs roll out, and the speed of the Bitcoin network increases, the protocol governing how much computing you need to earn a bitcoin responds by increasing the difficulty of the algorithm, meaning that it takes more hashes on average to create new coins. In the four years since Bitcoin was created, miners have repeatedly upgraded to devices that are more energy efficient and tuned to the task of calculating Bitcoin hashes, graduating from CPUs to GPUs to FPGAs, and now finally to ASIC mining rigs. Because each improvement has brought with it an increase in difficulty, first-generation machines that once brought in a profit now fail to mine enough bitcoins to pay for the electricity running them. Satoshi Nakamoto, the mysterious character who invented Bitcoin, most likely mined the first coins on a personal computer. Today, it would be absurd to do so. Perry now owns multiple ASIC miners, and when he put the first one online this April, he retired an entire rack of obsolete rigs.  

This trend is already pushing out some of the early Bitcoiners who can’t afford to upgrade. But what does the future look like? Perry is optimistic that bitcoin mining won’t be restricted to the richest: As rig manufacturing becomes a more professional enterprise, companies will find a way to bring a more diverse selection of products to market, bearing a range of price tags, he says.

“Imagine you made a low-power chip, something that costs a few cents to produce and sucks down no more power than the Wi-Fi on your phone. You could put that in everything. Put that in phones, TVs, laptops, tablets, etc. ‘The Internet of Things,’ as the media are so fond of calling it,” says Perry. In that scenario, everyone is mining for bitcoins, and although they might be collecting mere pocket change, they are still helping to maintain the distributed nature of the network.

But there is another way out of this arms race, or at the very least a way to slow down the impact that ASICs are having. Right now, the alternative is playing out as a separate hybrid cryptocurrency called Litecoin.

In 2011, after identifying a couple of potential weaknesses in the Bitcoin protocol design—including slow transaction times—a developer named Charles Lee altered it slightly and started a new similar currency called Litecoin. “The design idea was to create a silver to Bitcoin’s gold,” says Lee, who recently signed on to develop Coinbase’s online Bitcoin wallet while continuing to work on Litecoin.

The most controversial modification Lee made to Bitcoin was replacing the hashing function with a more memory-intensive cryptographic algorithm called Scrypt.

10NWBitcoin f4 Litecoin Hashrate

10NWBitcoin f5 Litecoin DifficultyLitecoin lifts off: The hashrate for Litecoin, an alternative currency, has increased as obsolete bitcoin-mining computers are repurposed for Litecoin. But older computers can compete on this network, because the repurposed miners don't dramatically affect the hashrate and hash difficulty.Charts: BitInfoCharts

Scrypt hashes are similar to the SHA-256 hashes that guarantee the irreversibility of Bitcoin transactions transmitted across the network. But finding a hash in Scrypt “requires a certain amount of memory, and requiring memory makes it hard for any chip to parallelize a lot of work,” says Lee. “In SHA-256, you can do thousands of hashes in parallel if you have a multiprocessor machine. But because Scrypt requires memory, and you only have a fixed amount of memory on your computer, you can’t do a lot of Scrypt hashes at the same time,” he explains.

As a consequence, if litecoin miners were to switch over to ASIC-enabled rigs, they would not see the same increase in profits that bitcoin miners have. Litecoin has therefore been called an ASIC-resistant cryptocurrency, because it reduces the economic incentive to upgrade.

“If you have a bitcoin ASIC machine, you can mine 50 times more bitcoins that way. It pretty much pushed all the GPUs out of the market because it raised the difficulty by 50 times right away, and GPU mining became unprofitable,” says Lee. “The difference between that and Scrypt is the ASIC Scrypt mining won’t be 50 times more efficient than GPU Scrypt miners. It will be faster. It will be a little more efficient, but not drastically more efficient like it is for Bitcoin.”

So far, investment in litecoins, which is now trading around US $2, has not been profitable enough to motivate any companies to develop an ASIC Scrypt miner. That could change, however, if the price of the currency goes up.

“It will cost millions of dollars in initial research and design and manufacturing cost. So if selling Scrypt ASICs will not recoup those investments, then it’s not worth it for someone to invest those dollars,” says Lee. “It will happen when litecoin value rises and there’s a bigger market. But right now, litecoins are too cheap for it to be worthwhile, at least not for another year.”

Until Litecoin gets sucked into an arms race of its own, the cryptocurrency may provide a good second life for all the older machines that are being taken off the Bitcoin network. Graphs of the mining difficulty on Litecoin over time show a huge jump in the processing power of the whole network occurring at the same time that bitcoin miners were unpacking their first ASIC rigs. This suggests that at least some of the bitcoin miners who upgraded may have put their old computers to work mining litecoins. And for everyone who wasn’t able to upgrade, Litecoin is becoming the next best option.

“After ASICs came out, the Bitcoin difficulty shot up so high that it became less profitable to mine bitcoins when compared to litecoins. So they all switched over,” says Lee. “For example, right now you can make more than six times as much money mining litecoins as bitcoins with GPUs. So pretty much every single GPU [mining rig] is mining litecoins now.”

But allegiances run deep in the world of cryptocurrency, and many have an attitude of “Bitcoin or bust.” When asked if he ever considered putting his old machines to work on the Litecoin network, Perry simply says, “Nah, I sold them a while back.”

One thing keeping Bitcoiners from taking Litecoin seriously is that none of the major online currency exchanges support it. This may soon change. Lee has spoken to the people running two of the largest exchanges—Mt. Gox  and Bitstamp—and reports that they are seriously considering adding them to their list of traded currencies.

About the Author

Morgen E. Peck is a New York City–based reporter who has been coveringBitcoin and other cryptocurrencies for IEEE Spectrum since 2011.

The Conversation (0)

Quantum Error Correction: Time to Make It Work

If technologists can’t perfect it, quantum computers will never be big

13 min read
Quantum Error Correction: Time to Make It Work
Chad Hagen

Dates chiseled into an ancient tombstone have more in common with the data in your phone or laptop than you may realize. They both involve conventional, classical information, carried by hardware that is relatively immune to errors. The situation inside a quantum computer is far different: The information itself has its own idiosyncratic properties, and compared with standard digital microelectronics, state-of-the-art quantum-computer hardware is more than a billion trillion times as likely to suffer a fault. This tremendous susceptibility to errors is the single biggest problem holding back quantum computing from realizing its great promise.

Fortunately, an approach known as quantum error correction (QEC) can remedy this problem, at least in principle. A mature body of theory built up over the past quarter century now provides a solid theoretical foundation, and experimentalists have demonstrated dozens of proof-of-principle examples of QEC. But these experiments still have not reached the level of quality and sophistication needed to reduce the overall error rate in a system.

Keep Reading ↓Show less