ASICs versus GPUs (Part 3)

Asics vs GPUs 3

Asics vs GPUs 3


In the “Two sides to every story” series, we will examine current controversies and issues in the blockchain space. Instead of choosing sides, we will do our best to present information from both (or more!) sides of the argument. Then, you decide!


In this series, our mission at CryptoMurmur is to provide you with information to help you sift through the facts and make your own decisions about where you stand on a topic. In Part 1 of the ASICs versus GPUs story, we examined the different tools that are used for the purpose of cryptocurrency mining. In Part 2, we examined the controversial implications of this dilemma in regards to the issue of centralization versus decentralization. Today, we look at hashrates, 51% attacks, and what role ASICs and GPUs play in regards to this.


Higher hashrates


When a Proof-of-Work network is more decentralized, it is more secure, since it does not have a single point of vulnerability. On the other hand, it is also slower and requires more hashing power as the entire network must be updated constantly to operate in a decentralized fashion. Theoretically, the more decentralized a PoW network is, the more secure it should be, but it sacrifices speed in exchange for this safety. The network needs to perform constant computations or hashing to verify transactions and to maintain security. So miners are incentivized to increase their hashing power for more frequent block rewards, which has the benefit of lending more security to the network. It is designed to be mutually beneficial: The higher the hashpower you have, the higher chance you have of being rewarded, and the more hashing power the network enjoys.

Additionally, networks that require higher hashrates — since there is more competition in the mining process — are more secure since they need more computing power to process, and therefore are more costly to mine. The higher the hashrate that a particular network requires, the more costly it is to set up or buy the necessary hashing power to launch what is called a 51% attack. At some point, it becomes so costly that it just isn’t worth attempting such an attack as the reward is less than the cost of buying the hashing power.


51% attacks


If any single entity or group of colluding entities manage to control 51% of a network, lots of bad things can happen. Most importantly, the 51% controlling entity can essentially double-spend the currency.

Here’s a bit of an explanation of double-spending: the 51% attacker sends currency on the consensus chain to be cashed out somehow (usually via an exchange, but it could be simply by buying something with the currency), then rolls things back by harnessing their superior hashing power. The malicious actor can use their higher hashing power to create the longest chain on a false series of blocks where these transactions were not completed, broadcast this false version of the chain to the network, and then make off with the money, thereby double-spending the currency. Some lower hashrate PoW networks like Bitcoin Private and Bytecoin are especially vulnerable to 51% attacks because it requires relatively little hashing power to take over these networks.

ASICs can contribute to centralization and the threat of 51% attacks if a few wealthy and powerful parties manage to gain more than 51% of a network’s hashrate. As mentioned before, Bitmain and some of its affiliates control nearly 50% of all the Bitcoin network’s hashing power — some say they already have more than 51% if you include some of Bitmain’s partners in the mix. Of course, the argument is that it would not be in Bitmain’s best interests to diminish the value of the Bitcoin network since they have so much invested in it. Yet, there is a degree of trust that is necessary because of the extent of their influence in the present conditions.


No going back…


It’s true that at least in the case of Bitcoin itself, there’s simply no going back to GPU mining. ASICs are here to stay and they do successfully contribute massive hashing power to the network, maintaining a much higher degree of security than other currencies — assuming you trust the controlling entities at the ASIC farming helm. To be fair, there are more ASIC competitors arriving on the scene. Most notably, Samsung is in the process of making ASICs and another company, GMO Internet Group, claims to also be planning to give Bitmain a run for their digital money. There are other competitors on the scene as well, but they have not made much of a dent in Bitmain’s monopoly thus far. ASIC competition could be good in the long run for the security of bigger blockchains like Bitcoin, but it might shut the doors on the average retail consumer being able to participate in the mining process of the higher hashrate coins.

On the other hand, ASIC-resistant PoW coins, mined by GPUs, remain quite vulnerable to services like NiceHash, whereupon one could buy hashing power sufficient to launch a 51% attack against these less secure, lower hashrate networks. Have a look at this article to see what I mean about how cheap it is to launch a 51% attack against some PoW currencies. A relatively low-hashrate GPU-dependent network simply does not have the necessary horsepower to stand up and retain hashing security against such potential exploits.

So again, it ends up being a bit of a toss-up. It’s true that on smaller, lower hashrate networks, ASICs or services like Nicehash can really tilt the playing field and over-power a network relatively easily. On larger networks like Bitcoin, Ethereum, and Litecoin, this isn’t really the case, although an argument could be made that they already are quite centralized and are therefore vulnerable to a few central entities controlling the networks — maybe not now, but perhaps some time in the future.

To conclude this series, next time we will examine what various cryptocurrency developers are doing in response to ASIC developments and in the battle, in some cases, to remain ASIC-free.




Leave a comment

Your email address will not be published. Required fields are marked *