or, is it better to use many small bitcoin solo miners ? ( which we know as lottery mining )
Some rough numbers for the second part of the question, since nobody put them side by side:
- A GPU running SHA-256 does only a few GH/s. A small SHA-256 ASIC like a Bitaxe does roughly 1 TH/s on about 15-20 W. That's several hundred times the hashrate for a fraction of the power, so for SHA-256 there's no case for GPUs at all.
- But even a Bitaxe on Bitcoin is a lottery ticket. With the network at 900+ EH/s, a 1 TH/s device has an expected time to a block in the thousands of years. Ten of them only divides that by ten.
So if the goal is income, it's an ASIC farm with cheap power or nothing. If the goal is a cheap, low-power lottery ticket, small miners are fine, and a few things improve the ticket:
1. Pick a pool whose fee and payout you can verify. Some solo pools take 2%, and as FP91G says, check whether tx fees are passed on too.
2. Smaller SHA-256 chains have much better odds per TH/s than Bitcoin, because the same hashrate is a far bigger share of the network.
3. Merged mining (AuxPoW) lets the same shares also find blocks on chains like Namecoin, at no extra power.
Disclosure, since it's directly relevant: MergedHash (mergedhash.com) is a non-custodial 1% pool that does points 1-3, i.e. BTC and eCash (ECX) solo or PPLNS, merge-mining Namecoin, Fractal and Hathor, with payouts in the coinbase. It's still luck, though. Run the numbers for your own power cost before buying anything.
Many modern graphics cards will not provide a hashrate exceeding 1000 megahashes, therefore, a miner would most likely mine profitable altcoins while receiving payouts in Bitcoin, there are plenty of services that facilitate this. Engaging in lottery mining with a hashrate of 1000 megahashes is a highly inefficient use of a graphics card's resources, though it might be worth trying as an experiment for a few hours or days.
