Quote from: pbfarmer on April 28, 2019, 07:59:00 AM

Quote from: NCarter84 on April 27, 2019, 11:33:50 PM

I've been killing myself over this one... Maybe, someone else from the outside can help me...

I recently upgraded my ATX 1000w PSU to a 1200w server PSU/breakout board. Upon that upgrade, as I was applying my overclocks and such through a bat file with ODnT, my system started freezing. If I applied this settings manually, no issues.

So, I did what any good miner would do.. DDU and start over.. Same thing... I DDU'd a 3rd time and upgraded to the latest drivers along with ODnT 2.8. Since then, my system hasn't had any issues applying the clock n suck through a bat file.

However, my hashrate that the miner shows vs what the miner shows pool side has been off... I've tried different settings in ODnT, lowered memory, more power and none of my tests are matching up..

I will run the miner for 12 hours, get both miner/pool hashrate to match up at 5.5 khs, stop miner, start again and run for 12 more hours and then they aren't matched anymore... 5.5 vs 5.3.... Sometimes the same test will result even worse at 5.2 or 5.1 -- I understand luck is in play, but I don't think luck should effect it this much.

My other rigs show Miner and Pool to be very close..

I have fixed diff at 55000 on Support XMR.

Moddest overclocks.

570s run 1150/875 and memory 1925/875.
580s run 1200/900 and meomry 2000/900.

I'll run the same clocks back to back and get different reported values each time... There are times where I have cards running 300-400 hs less than expected on the pool

I've tried more power, less memory, more memory...

I'm at a loss..

First, your pool and miner rate should technically not match. There's a 2.5% fee on CNR, so 5.3 is actually closer to the proper pool rate than 5.5.

Second, 12 hours is generally not even enough time to have confidence in a stable rig. It's definitely not enough time to get anywhere near a high level of confidence in an actual representative pool rate - you simply aren't submitting enough shares to make a proper measurement in that short of time. By my math, you would need more like 5 days for a ~99% confidence in your reported pool rate being truly representative, at your h/r and diff.

You actually nailed it on the head.... After doing the math, I forgot about the dev fee and was in panic mode because numbers weren't matching up... After reviewing all the test data, Every run was at 97% or better return hashrate except 1...

Thanks!