As Ethereum mining has increased in popularity, certain graphic cards/GPUs are becoming harder and harder to find. All of the RX460 and RX470 cards have been sold out for months, and if you do happen to get one, it will be for well above retail price.
That’s forced ethereum miners to look for other GPUs for mining rig builds. The GTX nVidia cards didn’t use to be worth it, but as the price of the Radeon cards has been driven up, the nVidia cards are now turning into popular options as well. That’s why I opted for the GTX 1060 cards in my mining rig. The real beauty of the 1060 is that after adjusting the overclocking settings and power target options, the 1060 can mine about 95% of it’s max hashrate at about 70% of the power usage – leading to a better return on investment.
Here’s the settings and results I got from my two EVGA GTX1060 6GB SSC gpus. For each setting change, I let them hash for an hour and then took the resulting average hash rate, and recorded the power usage from my Kill-a-watt power meter (we review the best power meters here).
To start, I increased the memory overclock settings until the GPUs weren’t stable for ethereum mining, and then I played with the core overclock setting (which didn’t seem to really change anything for power usage or hashrate). Lastly, I decreased power settings until my GPUs were mining ethereum at pretty much full speed with way less power consumption.
|Overclock Settings||GPU 1 Hashrate (mH/s)||GPU 2 Hashrate (mH/s)||Total mining rig power draw (including motherboard/power supply etc)|
|Stock, straight out of the box||19.037||19.108||215 W|
|Memory + 750 mHz||22.712||22.846||245 W|
|Memory + 1000 mHz||crashed||crashed||at this setting, the cards weren’t stable enough to run for an hour|
|Memory + 850 mHz||23.357||23.361||247 W|
|Memory + 950 mHz||23.774||23.984||252 W|
|I added core -100 mHz, + 100 mHz and no measurable difference was noted in hashrate or power consumption|
|Memory + 950 MHz, power target 70%||22.8||23.9||225 W|
|Memory + 950 MHz, power target 50%||22.7||23.6||190 W|
|Memory + 950 MHz, power target 45%||22.8||23.0||182 W|
|Memory + 950 MHz, power target 40%||19.3||18.5||140 W|
The sweet spot for these cards was with a memory overclock setting of + 950 mHz, and a power target of 45%. At this setting the power usage was around 70% of the highest hash rate, while hashing at about 95 % of the highest hash rate!
What I did notice is that my first card’s hashrate was always slightly lower than my second card. When I started looking at the info presented by Claymore’s miner, I saw the temperature on my first card was always 5-10 degrees hotter than the second. I finally figured out that since I had plugged these 2 GTX 1060s right into my motherboard, one of them had more airflow than the other – can you guess which one?
I have some PCI-e risers on order so that both cards can be up, away from the mining motherboard, and then they’ll hopefully run about the same temperatures and hash speeds.
Do you have 1060s in your Ethereum altcoin mining rig? What settings do you have them at?
P3 Kill A Watt Electricity Usage Monitor
This is the most widely known best energy monitor on the market. The Kill-A-Watt electricity monitor first came on the scene when the manufacturer published information on phantom power. When building a mining rig we don’t care about phantom power – we don’t every want our rigs turned off, so we are just concerned with power draw when mining.
Using the kill-a-watt electricity usage meter we can get a reading on how much power we are using to calculate the cost vs profit of our mining rig.
This power meter doesn’t just show you the power being used, it actually tallies the total amount used over a period. So, for example, you could start your mining rig, reset your power meter, and then check back in a week to see the total power used. Compared to the amount (and value) of ETH coins mined you can calculate your profitability.