Topic: Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v15.0 (Windows/Linux) - page 963. (Read 6590565 times)

Hi everyone
I have been a luker and this is my first post and contribution to mining community.
I found the problem and solution to this.
Check your  GPU Z sensors that your memory controller load is fluctuating and the hasrate is fluctuating in exactly same manner. (This was observed by few other people too here)
Now observe the GPU power and it fluctuates same way and goes to maximum of 110 W for my RX 470 Red Devil OC by Power Color. I read in a Overclocking forum that if you max out your power, your memory load will reduce. So I reduced my voltage through waltman and viola the fluctations reduced to almost minimal even at dcri 30 as long as my GPU power was around 100 W.

I am very new to mining (started last week) but I think that the issue that dual mining with ASM takes a lot of power and we know that RX 470 has a max tdp of 120 W. However, GPU Z reading of 110 W probably means that from the wall it is 120 W so GPU is throttling or something.
My system specs:
RX 470 OC Power Color Red Devil (Got them for 200 CAD dirt cheap and comes out of box frequency of 1270 and memory 1750) Samsung memory
BIOS stock (tried to flash BIOS but was unsuccessful).
Using AMD watmann I reduced stock frequency to 1200 and increased memory to 2100. Core voltage at 900,950 and memory voltage at 950.
-dcri 30
GPUZ: 100 W, 91ish % memory control load, temp is around 70 for 16x pcie card and 60ish for 4X pcie card.
Dual hasrate: ETH:26mh/s DCR: 800mh/s I know that my ETH hashrate is not as high as other people like 28+mh but I am still learning so who knows.

tl;dr Hashrate fluctuations with ASM mode are due to GPU hitting max power so decrease clock frequency and core and memory voltage to get a stable hashrate.

p.s  just wants to say thank you @claymore. Great work buddy.

I can confirm in linux (6xRX480) with powertune 7 (max powerlimit) the fluctuating with v9.0 and asm 1 is the same thing, unfortunately not fix the issue. Back to asm 0 which work just fine.

Hello
Which pool you advise me to decred?

I would like to teter eth + dcr, but I do not know many dcr pools.

thank you

I reverted back to -asm 0. Many of my cards (rx470 and rx480 modded bioses from all brands) were fluctuating. This resultted in about %10 decrease in effective hashrate.

Increase power limit doesnt work.
Higher or lower dcri reduces fluctuating (doesnt end it). But my profits are reduced compared to -asm 0.
Higher cpu priority had no effect.

So i m actively monitoring this thread to increase my decred hashrate %60 again 

core 1100, memory 1900.
Afterburner settings : -90mv and power limit -45

Try to see GPU load if you using MSI afterberner.
On multi GPU system with cheap 1-2 core CPU i noticed what this occurs.

Solution is to set in bat file
start /high EthDcrMiner64.exe #####blablabla####




or via task manager manual set priority to high

Try this maybe it help in your case.

!!!!!!!
-For everyone saying their hashrate is now fluctuating with v9.0 and asm 1.. try increasing power limit on your cards. at least like 10-15% and let me know if it makes a difference



I mine there too and that is normal for me.. i don't know if other eth pools are better


Same for me, usually at least around 1% a few of my gpus are more like 3% tho (maybe have too high OC or undervolt on these)


You want to undervolt to lower the energy consumption. Somewhere between 900mv to 950mv should work well. maybe even 880mv if you're lucky. Make sure you reduce both core and mem voltage to same value too


Try increasing power limit on those. if that doesn't work, try playing with clocks, voltages, gpu slot, riser, bios

This is the first rig, when starting and nancy, then one of GPU power down





This and rig2, and the same thing happens





This and rig3, all right, to be noted that the GPU are all the same, with the memory samsung 470 8gb, with strap 1625 to 2000.

risers are all 007 version sata attack

I can not figure out how to fix this

core/memory settings please?
you have 1x gpu more than me and same wattage, so probably  I set something wrong

I have corsair rm1000i and it takes 980w from wall. 6 x rx480, dualmining with 8.1 Consumption is ok, miner is 100+ hr stable.

There is something wrong with the new version (kernel), use -asm 0. I am doing dual ETH+DCR very stable but using -asm 0.  Unfortunately @Claymore said that he didnt see this behaivor in his gpus, but there are many report of fluctuation in this thread with v9.

Command line does not work for what reason? R9 290x super heating using command -tstop 75 it for mining, what do I need to add in the command to return automatically after set temperature?

-r 1
-tstop 75
-tt ?

You are forcing max load limits on your PSU. Consider adding a second PSU.

it kills ether+dcr hashes bruttaly

i'm using dcri 30, ether is now 143 h/s, dcr 4300 h/s
https://minerstat.xyz/public/coinmaker.html

but i'm worried about wattage, it didn't crashed from morning (last 6 hours). Is it very bad for PSU to be on that much load? it goes from 960-980w

-dcri 23.

Can someone suggest me which core/memory settings is best for my rig?

modified bios - 1750 straps to 2000 only

5x480 nitro OC+
core 1100  , 1000mv
memory 2100, 1000mv
-dcri 35
- asm 1

eth - 135 mhs / dcr-4700 mhs

980w from wall - is this to much for my Corsair RM1000x  ?
which settings should I change to get lower energy consumption from wall?

I have the same problem, with windows, drivers 17.3.3
with 8.1 it was stable ( both rigs 8xRX480 in total )
with 9.0 ETH hashrate goes up and down, dcr is stable, nomatter the -dcri

What operating system?
In Windows:

start - run - shell:startup - ( make shortcut of your bat file there ).

Set voltage to 900mV.

Hello

What is the command line to put in the remote manager .bat so that it starts automatically with my saved configuration in option?

thank you in advanceBTC