The most well-known coin to implement Ethash is Ethereum (ETH), for which this algorithm was initially created.
Its most prominent design goal is to be ASIC resistant, preferring to accommodate GPU-based mining.
The algorithm uses a large dataset that is periodically regenerated, and slowly grows over time. It fits in the VRAM of a modern GPU. This element of its operation is derived from the “Dagger” algorithm, and the large, varying dataset contributes to ASIC resistance.
The other part of ASIC resistance is achieved through a derivative of the “Hashimoto” algorithm, which exerts a lot of memory intensity in terms of random-access reads from the dataset. The inputs from the dataset are run through a derivative of the “Keccak” algorithm to output potential block solutions. This is well suited to the large core counts in modern GPUs.
Miners commonly make significant memory clock frequency increases on their GPUs to achieve considerable improvements in Ethash hash rates. Power or temperature limits, along with manually-set high fan speeds are sometimes seen, to offset the associated heat output.
It’s common to see custom-built PCs specified with 6 or more high-end GPUs for mining Ethereum, with minimal emphasis on the other components, to get the most cost-effective mining farms.