FPGA Miners: The Next Battleground in the Crypto Mining Arms Race
The cryptocurrency mining space has been dominated by two types of miners—GPU miners and ASIC miners—each with their own set of pros and cons.
While GPUs remain popular, some cryptos have been dominated by ASICs, which are far more powerful and consume less electricity than GPUs. However, ASICs—by design—can only mine a specific algorithm, while GPUs have the flexibility to mine almost any coin.
The problem is if someone figures out how to use an ASIC to profitably mine an algorithm currently mined by GPUs, ASICs will soon dominate that network. When the number of ASICs mining a coin increases, its network hashrate and difficulty level both increase, essentially pushing GPU miners out.
Certain cryptos like Ethereum remain ASIC-resistant—which means nobody has yet been able to economically mine these algorithms using an ASIC.
Some cryptos like Monero have also resorted to hard-forking a coin to shake ASIC miners off its network and bring GPU miners back in. Although this tactic has worked, ASIC miners and their manufacturers have now wised up. Manufacturers like Bitmain can now adapt an existing chip to an upcoming hard fork and update their ASICs accordingly within 3-5 months.
These dynamics are contributing to an arms race for faster and more efficient miners. We can soon expect to see new miners nearly as powerful as ASICs, but also as flexible as GPUs, making them more resistant to hard forks. In fact, with technological advancements, a third type of miner—the FPGA—is beginning to make a comeback.
Key Metrics for Comparing Miners
To objectively compare between three different types of miners, we need to use some key metrics as a basis. While some metrics—coin price, mining difficulty, network hashrate, and cost of electricity—are beyond our personal control, others are specific to mining equipment.
- Power Consumption
- Hardware Cost
The most important factor is the hashrate-to-power ratio. The higher the hashrate and the lower the power consumption, the more profitable the miner.
The ROI and durability on mining equipment is also important—the higher the initial cost, the longer it takes to recoup all upfront expenses. A robust and durable miner with high uptime rates also makes it easier to recoup the initial expenses. With a more favorable hashrate-to-power ratio, a miner may see a positive return within a shorter time frame.
Although other factors do affect mining profitability, they are not miner-dependent, such as the coin’s price, level of interest, and difficulty. If the coin’s price goes up due to hype and speculation, then more miners join the network, increasing its hashrate. The difficulty then rises. It becomes harder and less profitable to mine a block. To further complicate things, cheap electricity is necessary to stay profitable while running the equipment.
Pros and Cons of Each Miner
GPUs are popular because they are affordable, easy to source from Nvidia and AMD, and provides flexibility to mine different coins. Because GPUs can be reused for other purposes like gaming and AI, miners can easily recoup their initial costs by reselling them at 20-50% retail.
However, they are much less powerful than ASIC miners. Plus, if a coin is already dominated by ASICs—as are Bitcoin and Litecoin—GPU miners are not profitable. GPUs also consume so much electricity that they need cooling systems to dissipate excess heat.
- Widely available
- High energy usage
- Not the most powerful
- Cannot mine certain coins
- Bulky and heavy
Having the most raw power, ASICs are by far the most efficient miners on the market. They are also the most expensive. ASIC miners are designed to do only one thing: mine one specific algorithm of a specific cryptocurrency. This can be both a strength and a weakness. Unlike a GPU miner where you can choose which coin to mine and when, the ASIC miner immediately becomes useless when a hard fork occurs.
For these reasons, buying an ASIC ties you to the fate of one specific coin in a volatile cryptocurrency market. Additionally, some manufacturers of ASIC miners wield a lot of power in this space and are also not known for paying fair.
- Low power consumption
- Highest efficiency
- Extremely high hashrate
- Can only mine one algorithm
- Tightly controlled supply
ASICs are best for miners with large budgets, access to very cheap electricity, and supreme confidence in the future of a particular coin that is not ASIC-resistant.
On one end of the spectrum are the most flexible but least powerful devices, like CPUs and GPUs. The other end has the most powerful but least flexible devices, like ASICs. FPGA miners lie somewhere in the middle. Although FPGAs are not as powerful as ASICs, their best advantage is that, unlike ASICs, they can be reprogrammed to mine a different algorithm after a hard fork.
However, reprogramming an FPGA takes a lot of work and technical expertise. It isn’t as simple as merely clicking a few buttons to get a GPU to mine a different coin.
- High hashing power
- Energy efficient
- Generates less heat and noise
- Not as powerful as ASICs
- Still need adequate cooling
- Difficult to reprogram
- Limited resale value
However, the new FPGAs coming this year are expected to be several times more efficient than most GPU miners currently on the market.
Why FPGA Miners Are Coming Back
However, it may soon be possible to use FPGAs to mine ASIC-resistant algorithms at a much faster rate than GPUs.
As opposed to ASICs being rendered useless after a fork, FPGAs can be reconfigured to mine the forked coin. As Tyson O’Ham says, “these FPGAs aren’t as efficient as purpose-built custom ASICs are, but it’s their flexibility that makes them nearly impossible to stop: all a user has to do is load a new program into their board, and start mining again. Voila.”
If this does pan out, GPU miners will become less viable. There may be a window of opportunity lasting 6-12 months where the profitability of FPGA miners could surpass most GPU miners on the market. This window of opportunity could last until most networks become saturated with miners switching to FPGAs from GPUs.