feel free to ask me rather than speculate i coded the damn miner and made the screen shot guy
Ok.
a) Why is one miner generating more than 10 times as many shares, at the same hashrate?
b) Why do these shares not seem to be leading to any found blocks?
That's both of the things we have been trying to find out, although the shares are leading to found blocks also, there seems to be a massive discrepancy in hash rate reported between local and coinex and number of valid shares submitted vs the regular minerd64_sse4. On the Crypto-expert pool, which has round stats, the invalid % is in the 0.1 range for thousands of shares, so it's not submitting worthless invalid shares, but is submitting quite a few more valid ones between rounds than the sse4 version.
My equipment is 2 Supermicro 2u units, one with dual 6 core westmeres (24 cores) and another dual quad core HT xeon (16 cores) and an i5 (4 cores)
on Coinex pool, All of that will break 30mhs, isn't right when locally I'm reporting 100-115(sse4)-150+(spoetnik's sse4.1)khs per core. The amount of blocks found at current diff though, is in line with what I should be finding.
He's not actively trying to rip anyone off, just trying to increase the speed of the miner, and we have gone through several version tests already, and the one time the pool share discrepancy way over paid me on 2 blocks, I tracked down the finder (swiftshoot) and gave him the coins, and moved the largest part of my rig over to coinmine.pl where the other big guys are to keep from taking more. It's not an easy problem to track down, 3 pools, 3 different results, and no easy way to test other than on a pool....
Hopefully CaptainFuture can find out what is causing the massive share difference at coinex and resulting mega hash rate difference.
