[latchkey]

Except they never switched.

For security reasons, there can only be one top coin per class of hardware.

sha256/bitcoin = asic

ethash/ethereum = gpu

randomx/xmr= cpu

Yes, there are ethash asics, but they are effectively just asic gpus with ram... the memory controller is the gating factor because ethash is memory hard [1].

The gpu balance will shift with the merge... all the hash will go to ETC and other shitcoins. $21m a day in rewards will go to $1.2m a day.

A lot of GPUs will be turning off as the profitability drops. As profit drops, large miners will sell to the retail market to cover their costs...

ETC will trend towards zero... miners will try other coins, but those will also trend to zero since they have no actual use (utility) other than speculation.

These next couple weeks are going to be fascinating to watch. This merge is not only the end of ETH mining, but it could also be the end of speculation profit for a lot of other (shit)coins.

[1] https://www.vijaypradeep.com/blog/2017-04-28-ethereums-memor...

[cowtools]

Interesting, thank you for clearing this up.

However, I think you could have multiple competing cryptocurrencies in the ASIC class, because the ASICs are not as generic as GPUs or CPUs, so they cannot be easily repurposed for attack unless they are FPGA-based (doubful).

Anyways, I'm looking forwards to the new supply of GPUs. Hopefully we don't get the same thing again with Chia hogging up all the storage on the market or Monero hogging up all the CPUs.

[latchkey]

Correct, asic's could have multiple, but in reality, don't. There is litecoin with scrypt. I was the 3rd largest scrypt miner there for a while. But these are all speculation coins without any solid utility. Nobody is going to build asics for coins that don't have demand. We saw this happen with Grin coin.

Another little tidbit for you... GPU mining didn't really impact the supply as much as people like to talk it up in the press. Maybe for a short while, but it was more just a lack of manufacturing than anything.

Because ethash is memory hard, older (4-5 year old) gpus are actually the most ROI profitable. It isn't the same nm hardware race as asics. People buying up the most expensive GPUs for mining were not helping themselves at all.

CPU based mining is kind of a shit show... less about people buying CPUs and more about bot networks being used to mine.

[tromp]

> But these are all speculation coins without any solid utility. Nobody is going to build asics for coins that don't have demand. We saw this happen with Grin coin.

Not only is Grin the one coin, out of many thousands, that strongly deters speculation with its pure linear emission, but it did in fact have ASICs built for it (the Ipollo G1/G1-mini).

[latchkey]

Heh, hi tromp, you must have a grin filter? ;-)

I still stand by what I said... grin doesn't have any demand (the price is in the dumps and it has no utility)... nobody is going to pay $9k for an asic to earn pennies. how many of those things have people actually bought?

[tromp]

Despite its anti-speculative emission, Grin still has more demand than all but 25 PoW coins [1] and a G1-mini ASIC miner costs as little as $289. Since 1 miner does about 1.2G/s the overall graph rate of 8830 G/s suggests about 7360 of these have been bought.

[1] https://www.f2pool.com/coins

[latchkey]

No wonder the price is so low... all the miners just dumping on retail in order to make their capex back.

I appreciate that you tried to make a coin and that your algo is actually pretty interesting... but it is also too bad you didn't do a memory hard algo and be in a position to capture the 18m+ GPUs that are about to need a new home.