Everything used to be so simple in the crypto mining world: download the source code and let your processor do the work. Anyone with a computer at home could get in on the action and there was money to made for everyone, especially those lucky people residing in a country where electricity comes cheap.
Well, things have changed now, and many would say it’s not for the better.
The original vision for the bitcoin network – and the same goes for the later Ethereum network – was for a decentralised network staffed by the average Jack or Jill with their own device. That was underscored by a democratic vision where the bookkeeping of the network was distributed in such a way that it would be impossible for a single corporate entity to control the mining work that verifies and confirms transaction blocks.
It was not long before CPUs were eclipsed by GPUs and then, later, for ASICs – application specific integrated circuits – to push out GPUs, at least as far as the bitcoin network is concerned.
Now the Ethereum network, who’s architects were aware of the ASIC threat at the time of its conception in 2014, may fall victim to the centralising danger said to be inherent with ASICs. As a result, miners are on the lookout for ways to diversify GPU usage and there is a new option available in to artificial intelligence (AI) – it is the power-hungry space of graphical rendering.
The AI diversification route has been around for a while but rendering is relatively new, but new startups, such as Leonardo Render are aligning their business model and technology with the needs of GPU miners looking to repurpose surplus capacity.
So What Exactly is the ASIC Threat to GPU Miners?
ASICs, as their name intimates, are hardwired to work with the hashing algorithm of a blockchain’s protocol – they do one thing very well.
So successful have they been in the bitcoin sphere, it is now unprofitable to attempt to mine using a GPU. GPUs – graphical processing units – proved superior bitcoin miners to CPUs because they are able to process tasks in parallel which meant they could achieve much higher hash rates that with a CPU’s serial method of calculation.
However, in Ethereum mining the GPU is still dominant but that could be about to change with the first ASIC miners for Ethereum due to ship in July from Chinese mining giant and ASIC manufacturer Bitmain.
Bitmain announced back in April that the rumours it was working on an ASIC for Ethereum mining were true, although little is known about the real-world performance gains to be expected from mining with the Antminer E3, as opposed to the claims made by the manufacturer.
However, if the claims made by Bitmain are anywhere near accurate, the days of the GPU Ethereum miner may be numbered.
The new Antminer E3 is said to be able to perform at a hash rate of 180 MH/s. That compares with around 232 MH/s for an eight-processor rig of AMD RX580 cards.
With the Antminer E3 priced at $2,150 compared to the eight-rig GPU set-up coming in at around $2,800, the efficiency differential is substantial. When Bitmain starts shipping the E3 it could see a huge increase in the secondary resale market for GPUs as owners look to recoup some of the cost of their now surplus-to-requirement units.
Image credit: Wikimedia
Centralisation Dangers, Diversification Answers
Bitmain of course sells its ASICs to all comers so on the face of it there is no reason why there should be a centralisation of mining power. But Bitmain is also the largest end user of its product, with huge mining farms packed full of its ASICs.
Bitmain controls about a quarter of the hash power of the bitcoin network through its AntPool and BTC.com mining pools. With its sights set on the Ethereum network the question of diversifying GPU usage is now a pressing one for miners.
As we mentioned, the most prevalent alternative is in the artificial intelligence (AI) field, where processing power is in high demand and GPUs perfectly-suited to the task because machine learning in neural networks, for example, typically involves computing lots of small calculations where parallel processing works well.
Demand for computing power is being spurred on by growth in AI and that is set to continue as more and more applications are developed and the science in turn continues to improve, leading to yet more demand.
As we have seen there is also a second option, also derived from a new trend in computer usage, which is also expanding rapidly, possibly at an exponential rate, in the form of graphical rendering.
We almost take it for granted to expect to see seemingly realistic-looking spaceships engaging in inter-galactic battle or armies of warriors battling it out in a fantasy world. Such scenes are only possible because of the marshalling of huge amounts of computing power to render computer-generated images (CGI). The resource required is considerable. Using a Nvidia 1060 GPU it would take 24 hours to render one frame and there are 24 frames in a second of film. Incredibly, that means it will take a month to render one second of CGI animation.
With CGI techniques in demand in film, gaming and in the wider corporate world for communications and advertising, there is a bottleneck in the mismatch of demand and supply.
Graphical Rendering Needs your GPU Power
Leonardo Render, which already supplies rendering services, is pivoting, with the help of blockchain, to provide a decentralised graphic render solution that can bring supply into contact with the demand. It will offer its clients access to a blockchain render farm with the help of its partnership with US mining facilities provider Giga-Watt, which has 23,000 GPUs on 11 acres of land on the US West Coast.
The part of their offering that jumps out for crypto miners is the yield available from rendering. Graphical rendering sees an increase of 1,751% on the returns that can currently be made from mining on the Ethereum network – $0.027 compared to a Leonardo PPU miner at $0.50, measured in GPU/hour.
This feels like a very smart move by Leonardo Render, with a solution that will be in the right place at the right time as Bitmain starts to ramp up its Antminer E3 production lines and Ethereum’s GPU miners search out alternatives. Crypto miners will be able to provision their units on the Leonardo network through the rendering platform’s LEOS token.
The Ethereum ASICs from Bitmain are not out in the wild yet (although there are rumours it has been quietly mining with them unannounced), so the promised efficiency gains have not been put to the test. It will also be interesting to see how Bitmain has tackled the Ethash Ethereum algorithm’s requirement for large amounts of storage, forcing a miner to spend the greater part of its time reading from a 1GB dataset as opposed to performing computational tasks.
There is also the possibility that the core developers on Ethereum will come up with upgrades that could “brick” (make unusable) the ASICs.
However, there are many other blockchain networks that could be susceptible to ASIC mining in addition to Ethereum so the issue of a profitable alternative use for the fixed costs sunk into crypto mining investment is not going to go away. AI and rendering to the rescue.