someone from the community shared that using M3 Max made increments grow to 700,000 quickly and then converting it to 9654 got a high yield of 900-1800quil/day, this sounds over the top to the point of being cheating and the community is very vocal about it, are you aware of this? I doubt there’s miners have use it for mass-producing profits. It’s like a spot-check cheat. I don’t know the scale of doing this, imo switching from the same core to the same or lower core doesn’t count as cheating, switching to the higher core does.
If these are true, it would harm the honest users within the community. Does anyone have good solutions to deal with this situation?
I don’t know what the community members think about this. Please leave your opinions and suggestions .
If some malicious users first use high-frequency CPUs to quickly increase increments, and then build a cluster on this, would this result in malicious farming ?
It takes weeks to reach 700k increment even on an M3 Max and 900-1800 quil/day on a 9654 is unlikely even at that increment. Then there is the question of what to do with M3 Max: V1.4.21.1 Release Notes - #8 by abc
The config swap strategy is well known and people have been doing it since the start of .19. While I do not think it’s cheating, I do think it is rig optimisation - like you would overclock your CPU to get through increments faster (and thus more rewards).
Remember, all the rigs are still working for the network and contributing to it, even the increment blaster. If someone was to do this with some “offline” worker then I’d probably say yeah that’s not entirely right.
Those reward figures you’ve mentioned on a 9654 are not possible, even at 700k increment. A 9654 at 220k increment earns about 160-165 Quil per day in rewards, you can calculate the rest with those numbers.
For example, I have an AMD 7950x (16-core 4.5 GHz) and an AMD 9654 (96-core 2.4 GHz).
The increment on the 7950x will grow much faster than on the 9654. Assuming my 7950x reaches an increment of 500k, because the higher the increment, the faster the mining efficiency. However, the 7950x only has 16 cores, while the 9654 has 96 cores. So I packed the 7950x config and store file to the 9654 machine. This way, the mining efficiency would increase a lot.
Is this by design and acceptable ?
While you are doing this, you were earning very less coin on 7950x. Did you calculate yield of that situation? You will earn higher rewards for very short time.
It is not considered cheating to migrate hardware, whether it’s to faster, smaller core counts or slower, larger core counts.
That being said, the numbers are somewhat surprising.
At 700k increment, the difficulty becomes 25,000. On a 9654, this is roughly 2.5s. At 96 cores, 2.5s proof times, somehow waving away the KZG time as well, this would produce a rate of 95*(60/2.5)*60*24*1200000/8000000000 QUIL/day = 492.5 QUIL/day
If at 25000 the 9654s are actually doing better than 8x the rate they had at the start, that would be extremely interesting to see.
One thing that is important to know for folks machine hopping: if you try to claim two or more sets of proofs for the same peer id, the first one claimed is the one that will succeed, so please be sure you are not trying to copy the same data set to multiple nodes in the hopes this will double/triple/etc the mining rate.