2x GTX 1060 3GB - setting up my first miner

Hi there,

I’m new to mining, started about a month ago using my home PC with an GTX 1050 on board. Tried mining Zcash at first but then switched to some smaller coins.

Anyway, now I’m building my first mini-mining machine. So far got myself the following:
GPU 2x Asus GTX 1060 3GB (Hynix mem unfortunately)
CPU Intel G530T - 35W TDP
MOBO Asus P8H61-M LX R2.0 - I hope it will support 2x GPUs as I haven’t found too many information about this motherboard
MEM 4GB DDR3 1333MHz
1x USB riser 009s
and now looking for some low power ssd/hdd and a decent ~600W PSU.
If the motherboard will work fine with multiple GPUs I’ll consider adding another 1 or 2 GPUs in the future.

I’m planning to use EWBF miner - is it a good choice?
Yesterday I’ve been working on one of the GPUs settings to find the most efficient configuration and so far I got such results:
Default (Power Limit 100, Core/Mem +0/+0) - about 246 Sol/s, 100W, 2.46 Sol/W
OC (Power Limit 100, Core/Mem +150/+500) - about 274 Sol/s, ~102W, 2.69 Sol/W
Efficient (Power Limit 60, Core/Mem +150/+500) - about 238 Sol/s, ~70W, 3.4 Sol/W

What do you think about those results? Are these fairly OK considering current difficulty level and such?
I’ve seen 1060s getting to 4 Sol/W but those were topics/videos from ~6 months ago or older so I’m not sure if I should compare my results to those.

i do above 300 hash/s at 85W core around 2030-2050Mhz Mem 4000Mhz Voltage 0.96V with gigabyte wind force (2 fans) 1060 3 GB

btw u dont need to push the mem so hard rather beck them down that will give u few milivolts that can be used to pump the core a bit that will do u better results
image

1 Like

I’ve tried pushing both the core and memory and it doesn’t go much further.
Core ends at about +220MHz @ 80% power limit (effective around 1900MHz), even with memory set to default. Above that it isn’t stable anymore.

sett the power limit so you have around 0.95 - 1.00 V on the core (each card is different some will give this voltage on 85% some on 90%) then you should be able to push it over 2000 (tested on gigabite , zotac , MSI 1060s )… u dont mentioned the voltage u have on the cards…btw if your Voltage is like 0.8-0.85V at around 85-90% of power limit you have some problem in power distribution (like sata power connector used for risers) or crappy riser (older versions with crappy stabilizers ) or possibly week PSU …

1 Like

I have a 1060 3gb.
Approx 300 sol/s, at +175 core and +25 memory with 80% power.
I’ve done lots of tests and found the best success with LOWER memory and MORE core.
Basically, throw what you can at the core, then if you have any headroom left, bump up the memory.

You can see it running at approx 3 mins in:

THIS THIS THIS x 1000

Everyone is like "push memory as hard as you can, pump memory hard… etc…"
I’m sorry, but I’ve done hours of testing and found the BEST sol/w is ALWAYS, push core as high as you can [generally get it up to 1.9+2.0ghz] before crashing, then allocate any extra power to the memory.
I can run memory at -400 and core at +180 and get better sol/s then memory at +100 and core +160.
SHA256 is not very memory speed dependant, or if it is, the base clock on most cards is fast enough as it without overclocking.

2 Likes

Sorry for the mess, I don’t know how to edit posts here properly.

These are very nice results you’re getting there.
It seems no matter what I try I can’t get past ~280 Sol/s.
I did as you suggested, dropped the memory to default and pushed the core as far as I could which is around +200-210 (at +220 already crashed once), which is about 1940MHz effectively and at 80% power limit (~0.98V) gives about 275 Sol/s @ 92W.

Could it be the drivers, OS or the miner/pool that stops me from getting that extra 10% hashrate?

BTW. I have a question, maybe silly, but does it matter if my motherboard only has USB 2.0 and the riser features an USB 3.0 ports and cable? Will it work just fine? I know the USB is only used to connect the two riser PCBs so I think the motherboard doesn’t matter but I just want to be certain I won’t fry anything when I put it together.

USB revision on motherboard dosent matter at all
the USB on the riser dosent work as a USB…it is just a connector to connect 4 cables(pins) that will carry the data to your PCIe…it could be any other connector type but since USB cables are cheap and already mas produced it is cost wise to use a USB plug
what matter is the generation used for the PCIe ports

these are mine specs ± on all the 1060

So I installed exactly the same drivers and basically copy-pasted your settings and here’s what I get:


so average of ~282 Sol/s (@0.98V).

Apparently my Asus 1060 has lower default core clock, but would 100MHz lower core clock mean losing 20 Sol/S? I think it has to be something else or am I just unlucky with the card itself?

Pushing it to +205 core and +400 memory get’s me only to ~284 Sol/s.

Well, I am using DSTM miner version 0.5.7, and you are using EWBF, that could the cause of the speed difference.
DSTM is faster, could get you to 300 and beyond.
Also, 1923 core is respectable. I’ve never got any card of mine stable past 2k, so props to koso for that [at 85% power no less]
I think if you change miners and leave settings the same you will see a large increase in speed.

1 Like

You were right! The DSTM is indeed faster, getting 300 Sol/s @ 83% power limit and +200/+400 clocks, or 266 Sol/s @ 60% power limit, giving me over 3.7 Sol/W, awesome!

Shame it doesn’t have any intensity feautre for my home pc, but for the miner it should do better than EWBF, as long as it will prove to be stable.

hey man u cant get more from the card most probably because u run it at almost 70 deg the best for these chips is to keep them under 60 ideal 50 …these pascal chips are like this the more temp you have on the gpu the less voltage it is giving to the chip this will auto down clock it no matter the setting u sett… but happy u hit at least 300 with DTSM :smiley: but settings gite high on OC

Been playing around with my 6x 1060’s today. and +500 on the Samsung ram. 90 watts power. Trying to find a core that holds. 1960ish holds and gives avg of 310 h/s. Some card in there is crashing it. Probably the hynix ram. Going to go through some long stability checking on each core to see where each card crashes. (Getting stability above 1960 with 90 watts power seems to be a ceiling, even if I drop the mem down to +400) using dtsm. Now 90 watts is 75% of the cards 120w tdp. Maybe at 100w I could get 320 h/s out of them.

@koso16 the temps will be lower as right now I’m just testing the card while it’s sitting in my closed case home pc, but this week we’ll be setting a rig in the basement which is going to be a much cooler environment.

Still I think I’ll run them @ 60-70% power limit as it’s the most efficient per W and electricity here is quite expensive too.

1 Like

ok go and install gpuZ and check the memory controller load in sensors tab this will show you how effective is your setting(how effective is work addressed on your Vram from you Gpu chip) it will also help you fine tune the memory to the max core frequency that you can get

ok so for a closed case system there results are quite ok :smiley: still u will need to fine tune them when the miner is in the basement

So the rig is finally running on Ubuntu linux, for the first couple of hours it’s been performing like this:

Safe OC settings of +180/+450 and power limit set to 68W. I will be trying out higher power limits later but I think this is a sweet spot in terms of Sol/W (currenly averaging 3.83).

I’m not sure why there’s such a difference in temperatures of both cards. Okay the GPU1 does have a little more restricted airflow (gonna improve that) around it but even with fan set to 70% it gets to 65*.
I don’t think it matters but GPU0 is installed on the motherboard and GPU1 via usb riser.

Overall I’m really happy with the results, especially considering how cheaply I managed to buy the whole system. Looking forward to adding more GPUs in the future :slight_smile:

You may run into stability problems with a +180 core clock and only 68w. Normally people have the 1060’s in the 80 to 90w range. Temps can vary even if they are the exact same card. Thermal paste could be bad on one or it could be fans and airflow. I have 6 1060’s 3gb. 5 run between 55 and 57 and one at 60. All same global fan setting. Let us know how your stability goes at 68w. Sometimes people don’t set the core overclock and just let the power determine the output. Guess you could try playing around. That’s how most people find optimal settings for their rigs. Good luck.

I’ve improved the airflow and now the GPU1 runs much cooler than GPU0 :wink:
In terms of stability I’ve had no issues so far at these settings. Keep in mind that lowering the power limit reduces the effective core clock as well, so running it at +180 at 68W means effective ~1600MHz, while at 80-90W it’d be around 1800-1900 I presume.

Good to hear that it’s running stable. Yes reducing power also can reduce core clock, just setting an offset and reducing power has caused me issues in the past. Especially if it close to What the card will take. Sometimes people just set the power and let it determine the core clock without putting in a + offset. But if it runs stable that’s great. What sols are you getting now at that power level?