Hi @tromp I’ve been reading a lot about Grin mining but there’s a question I haven’t found an answer for: how does Grin deal with mining difficulty and block production? Is there a automatic adjustment as in other PoW cryptos? Does it work the same way, or not, for C29 and C31+? Thanks in advance for your answers!
Looks like it was just a rash statement from their (Sapphire) side. On the wave of HYIP.
Right now 2080ti makes 1.7gps with fidelity of about 0.7. Does it mean that “real” useful gps is 1.7 * 0.7 = ~1.19, or that 1.7 number already has 0.7 fidelity applied?
All the GPS being reported assume fidelity=1. Currently 2080ti is around 1.25 effective, but @tromp is working on a fix.
I’m considering building a rig with 6x2070. 2080Ti are extremely expensive for me to take a risk, and 2080 costs 1.37 times more than 2070, but gives only 1.27 times more GPS…
What will you guys say about it? Should I choose 2070 or may be I missed something?
@Rabinovitch you also have to consider the electricity cost. 3x 2070 (@.5gps each) should get you graphrate of 1x 2080ti (@1.5gps each). Is 2070 1/3 of the price and does it consume 1/3 of the electricity? From what I can see 2x 2080ti should be same graphrate and less expensive and consume less energy than 6x 2070 .
Wow, thanx, I had no performance values for 2080Ti on C31 until now. ))
@tromp can you confirm fake 1.7 gps numbers for 2080ti right now on grin-miner? And effective 1.2-1.25?
where is the source for this quote? thanks!
is this means the real GPS for 2080ti C31 is 1.7*1.25= 2.125 ?
Read the whole thread. I meant, REAL GPS for 2080Ti is 1.25 gps. That’s an assumption, based on fidelity of 0.7, and needed to be verified by @tromp himself.
Minerbabe… v4.12, latest kbminer version.
1.7 sounds about right and it’s know to have fidelity around 0.7, so that has the same solution rate as 1.19 gps with fidelity 1…
Could someone explain what fidelity means? Does it mean a fidelity of 0.7 only solves it correctly about 70% of the time. thus the handicap.?
Is there a potential fix for this, or is this is just the way it is?
Fidelity is explained here: What to mine? Choosing between Cuckatoo31+ and Cuckaroo29
The quote is:
Fidelity of a miner is the probability of it finding 42-cycles in graphs that have them. Normally that should be over 0.999, so we can consider it equal to 1 for all intents and purposes. But when a (mean) miner is constrained by available memory, it has to spill a small fraction of edges that do not fit into the buckets and this reduces the fidelity. Notably, the cuckatoo_mean_cuda_rtx_31 miner has a reported fidelity of about 0.7
Thanks, that clears it up a little, so i’m guessing there is no fix for this card.
May be we will see such ‘fix’ in coming weeks. @tromp is working on an update with more relaxed memory constraints which can result in fidelity closer to 1.
that would be great, thanks
Taking Tromp’s and Greggy’s comment about .7 fidelity in mind, Sapphire’s RX570 0.42 gps effective becomes 0.6 gps (ie. 1/.7*0.42) or about 1/3 of 2080ti performance but about 1/4 the price.