What is Ether
Ether is a necessary element — a fuel — for operating the distributed application platform Ethereum. It is a form of payment made by the clients of the platform to the machines executing the requested operations. To put it another way, ether is the incentive ensuring that developers write quality applications (wasteful code costs more), and that the network remains healthy (people are compensated for their contributed resources).
How are ethers created
The total supply of ether and its rate of issuance was decided by the donations gathered on the 2014 presale. The results were roughly:
- 60 million ether created to contributors of the presale
- 12 Million (20% of the above) were created to the development fund, most of it going to early contributors and developers and the remaining to the Ethereum Foundation
- 5 ethers are created every block (roughly 15 seconds) to the miner of the block
- 2-3 ethers are sometimes sent to another miner if they were also able to find a solution but his block wasn’t included (called uncle/aunt reward)
Note that after the Byzantium update is implemented, the mining and uncle reward is reduced to 3 ethers and 0.625-2.625 ethers, respectively.
Is the ether supply infinite
No. According to the terms agreed by all parties on the 2014 presale, issuance of ether is capped at 18 million ether per year (this number equals 25% of the initial supply). This means that while the absolute issuance is fixed, the relative inflation is decreased every year. In theory, if this issuance was kept indefinitely then at some point the rate of new tokens created every year would reach the average amount lost yearly (by misuse, accidental key lost, the death of holders etc) and there would reach an equilibrium.
But the rate is not expected to be kept: sometime in 2018-2019 Ethereum will be switched from Proof of Work to a new consensus algorithm under development, called Casper that is expected to be more efficient and require less mining subsidy. The exact method of issuance and which function it will serve is an area of active research, but what can be guaranteed now is that (1) the current maximum is considered a ceiling and the new issuance under casper will not exceed it (and is expected to be much less) and (2) whatever method is ultimately picked to issue, it will be a decentralized smart contract that will not give preferential treatment to any particular group of people and whose purpose is to benefit the overall health and security of the network.
How do I mine ether
The Ethereum network is kept running by computers all over the world. In order to reward the computational costs of both processing the contracts and securing the network, there is a reward that is given to the computer that was able to create the latest block on the chain. Every 15 seconds, on average, a new block is added to the blockchain with the latest transactions processed by the network and the computer that generated this block will be awarded 3 ether. Due to the nature of the algorithm for block generation, this process (generating a proof of work) is guaranteed to be random and rewards are given in proportion to the computational power of each machine.
GPU Ethereum mining
- Ethereum’s proof of work algorithm does not make use of Scrypt or Sha256, instead, it leverages EtHash, a Hashimoto / Dagger hybrid. You can read all about the theory behind this and its design in the Ethereum gitBook, mining chapter. Note that for Serenity (a future release, a major milestone on the Ethereum development roadmap) we are planning to switch to Proof of Stake (PoS).
- The Ethash proof of work algorithm is memory hard, you’ll need at least 1+GB of RAM on each GPU. I say 1+ because the DAG, which is the set of data that’s being pushed in and out of the GPU to make parallelisation costly, will start at 1GB and will continue growing indefinitely. 2GB should be a good approximation of what’s needed to continue mining throughout the year.
- Mining prowess roughly scales proportionally to memory bandwidth. As our implementation is written in OpenCL, AMD GPUs will be ‘faster’ than similarly priced NVIDIA GPUs. Empirical evidence has already confirmed this, with R9 290x regularly topping benchmarks.
- ASICs and FPGAs are strongly discouraged by being rendered financially inefficient, which was confirmed in an independent audit. Don’t expect to see them on the market, and if you do, proceed with extreme caution.