Quote from: xiggga on July 29, 2018, 09:24:28 AM

Quote from: akira128 on July 28, 2018, 09:48:23 PM

running ubuntu 18.04.1 with 13 RX570 (8GB) cards.

There is something seriously wrong with the Equihash 96.5 (Minexcoin) algo on v0.4
If I use the WORKBATCH: HIGH setting, the hashrate slows down to a crawl, also of note is that the power draw of each GPU is around 40 watts (it's normally 95 watts when mining). Something just isn't working right with the HIGH setting.
If I use WORKBATCH: MEDIUM, the reported hashrate (on http://eu.minexpool.nl) is around 10-15% higher then with v.034
However, it goes through these long periods where very little crypto is actually paid out from the pool. I don't understand how the listed balance for this rig is consistently really low, but yet the hashrate is high.

Switched back to v.034 for now.
v.034 doesn't have issues with '--set-work-batch HIGH ' , and it gets paid more from the pool than with v.04

the same thing Sad

I tried using v.041 today, but nothing has been fixed. Still having issues.
With 'WORKBATCH: HIGH'

-------------------------------------------------
   Welcome to lolMiner 0.41

   This software is only meant for pool mining
   with the Equihash 96.5 or 144.5 algorithm

   Made by Lolliedieb, July 2018
-------------------------------------------------
lolMiner is configured to mine MinexCoin(MNX), Equihash 96.5
Setup Miner...
Using device with id 0 (Radeon RX 570 Series)
Using device with id 1 (Radeon RX 570 Series)
Using device with id 2 (Radeon RX 570 Series)
Using device with id 3 (Radeon RX 570 Series)
Using device with id 4 (Radeon RX 570 Series)
Using device with id 5 (Radeon RX 570 Series)
Using device with id 6 (Radeon RX 570 Series)
Using device with id 7 (Radeon RX 570 Series)
Using device with id 8 (Radeon RX 570 Series)
Using device with id 9 (Radeon RX 570 Series)
Using device with id 10 (Radeon RX 570 Series)
Using device with id 11 (Radeon RX 570 Series)
Using device with id 12 (Radeon RX 570 Series)
Work batch: 28
Work batch: 28
Work batch: 28
Work batch: 28
Work batch: 28
Work batch: 28
Work batch: 28
Work batch: 28
Work batch: 28
Work batch: 28
Work batch: 28
Work batch: 28
Work batch: 28
Connecting to pool...
Connected to eu.minexpool.nl:3333
Subscribed to stratum server
New target received: 000028f5c28f5c28f60000000000000000000000000000000000000000000000
New job received: 1bdc8
Authorized worker: xxx
Start Mining...
Submitting share
Share accepted
Average speed (5s): 11906.4 sol/s | 11865.5 sol/s | 10575.0 sol/s | 10971.8 sol/s | 11924.7 sol/s | 10581.1 sol/s | 9939.8 sol/s | 10534.7 sol/s | 8552.7 sol/s | 9554.2 sol/s | 9892.5 sol/s | 10990.5 sol/s | 10571.5 sol/s Total: 137860.4 sol/s
Submitting share
Share accepted
Submitting share
Share accepted
Submitting share
Share accepted
Average speed (5s): 11904.7 sol/s | 11798.4 sol/s | 11482.3 sol/s | 11838.8 sol/s | 11863.1 sol/s | 11523.7 sol/s | 11147.2 sol/s | 11282.5 sol/s | 11169.1 sol/s | 10562.9 sol/s | 10235.7 sol/s | 11193.2 sol/s | 9182.5 sol/s Total: 145184.3 sol/s

It reports it's mining at 145 Ksols. However, according to the pool, the hashrate is much lower

Worker   Hashrate (Now)   Hashrate (AVG)   Diff.    Shares
rig1    34.36 KSol/s    50.72 KSol/s    -2457600   900

Please fix this issue! There's nothing in the logs to indicate there's any sort of issue. lolminer continues to report high hash rates.
..and yes, the difficulty value is -2457600. This number constantly changes to some random off-the-wall number.

Meanwhile the power draw of each of my cards are all over the place:

ROCm output with V0.41 WORKBATCH: HIGH
==================== ROCm System Management Interface ====================
================================================================================
GPU Temp AvgPwr SCLK MCLK Fan Perf SCLK OD
  1 49.0c 74.130W 1168Mhz 1750Mhz 16.86% manual 0%
  2 46.0c 71.100W 1168Mhz 1750Mhz 16.86% manual 0%
  3 48.0c 69.43W 1168Mhz 1750Mhz 16.86% manual 0%
  4 56.0c 68.216W 1168Mhz 1750Mhz 16.86% manual 0%
  5 48.0c 63.3W 1168Mhz 1750Mhz 16.86% manual 0%
  6 55.0c 68.94W 1168Mhz 1750Mhz 16.86% manual 0%
  7 42.0c 63.159W 1168Mhz 1750Mhz 16.86% manual 0%
  8 46.0c 62.99W 1168Mhz 1750Mhz 16.86% manual 0%
  9 51.0c 79.39W 1168Mhz 1750Mhz 16.86% manual 0%
  10 55.0c 69.3W 1168Mhz 1750Mhz 16.86% manual 0%
  11 59.0c 50.68W 1168Mhz 1000Mhz 16.86% manual 0%
  12 49.0c 47.175W 1168Mhz 1000Mhz 16.86% manual 0%
  13 44.0c 55.197W 1168Mhz 1000Mhz 16.86% manual 0%
================================================================================

**yes, the SCLK value is a little low, I've stepped in down a notch with:
rocm_smi.py --setsclk 5

However, even if I use '--setclk 6', I see similar disparities between the power draw and hash rates.

Also, the MCLK rates sporadically change between different values {300, 1000, 1750..etc} Mhz for each of the cards. Every time I run this output the MCLK values change.

**********update**********
I've been running V0.41 for the last hour or so with 'WORKBATCH: MEDIUM' And the results are pretty impressive so far. There's a definite 15-20% increase in the average hash rate over v0.34. Everything looks pretty solid so far. Need to run it for a few days to see if this trend continues.

**for comparison
ROCm output with V0.41 WORKBATCH: MEDIUM
==================== ROCm System Management Interface ====================
================================================================================
GPU Temp AvgPwr SCLK MCLK Fan Perf SCLK OD
  1 59.0c 91.180W 1168Mhz 1750Mhz 16.86% manual 0%
  2 55.0c 92.161W 1168Mhz 1750Mhz 16.86% manual 0%
  3 57.0c 89.70W 1168Mhz 1750Mhz 16.86% manual 0%
  4 67.0c 94.119W 1168Mhz 1750Mhz 16.86% manual 0%
  5 57.0c 88.94W 1168Mhz 1750Mhz 16.86% manual 0%
  6 65.0c 91.209W 1168Mhz 1750Mhz 16.86% manual 0%
  7 53.0c 90.72W 1168Mhz 1750Mhz 16.86% manual 0%
  8 56.0c 94.34W 1168Mhz 1750Mhz 16.86% manual 0%
  9 62.0c 97.141W 1168Mhz 1750Mhz 16.86% manual 0%
  10 70.0c 96.27W 1168Mhz 1750Mhz 16.86% manual 0%
  11 73.0c 97.100W 1168Mhz 1750Mhz 16.86% manual 0%
  12 61.0c 94.14W 1168Mhz 1750Mhz 16.86% manual 0%
  13 54.0c 91.69W 1168Mhz 1750Mhz 16.86% manual 0%
================================================================================

****************update 2*******************
After mining for a few hours, the balance that each rig below was contributing to the next 0.1MNX payout was roughly equal.
However, 6 hours later the hashrate on the lolminer rig was still high but is producing significantly less crypto.
The balance listed below between the two rigs should be roughly equal, but it's not even close.

Worker   Hashrate (Now)   Hashrate (AVG)    Diff.    Shares Balance
nvidia   167.23 KSol/s    167.67 KSol/s    1495173   547.55   0.04332093 (EWBF Cuda Equihash miner)
amd     157.71 KSol/s    147.99 KSol/s    1462857   714.29   0.01504778 (lolminer v0.41)

Why is this happening? V0.34 had a lower hashrate but was producing more crypto in the long term.
This is the exact same behavior I saw with version v0.40

**lolminer v0.41 did fail twice during the time I was mining. Both failures were due to: 'Floating point exception(core dumped)'
**I have a script that restarts the miner in the event that it fails. So these failures resulted in only a minute or two of downtime.

switching back to v0.34...