- Overclocking Nvidia GPUs under Ethash and Equihash
How efficiently your video cards work on the farm, directly depends on the profit obtained from mining. That's why you should squeeze the maximum out of the device. In this article, we will consider how to overclock Nvidia GPU and achieve the optimal mode of the video card when mining cryptocurrencies.
- OVERCLOCKING NVIDIA GPU: PREPARATION
Preparing for overclocking the video card should be at the stage of its purchase. Nvidia GPUs are equipped with different types of video memory, which include boards from the following manufacturers:
So, cards with Hynix video memory are not suitable for mining Ethereum and similar cryptocurrencies, giving a 20-25% slump in hash rate. But with such a card you can mine ZCash or HUSH. Video cards with memory from Micron and Samsung are more stable when mining ethereal coins. Determine what type of memory is on your GPU, you can in the properties of the device.
Now that we have clarified a little bit about how the seemingly identical video cards can differ, we proceed directly to the preparation of the software. For overclocking we will use the Msi Afterburner program. It is free and easy to use. Download the installer, run it, and wait until the installation is complete.
Go to the settings section, the tab "Basic", and check the boxes that are selected in the picture below.
Now everything is ready and you can start overclocking Nvidia GPU.
- OVERCLOCKING NVIDIA GPU FOR ETHEREUM MINING
We will be looking at how to overclock a graphics card using Inno3d 1060 6gb as an example but in fact, the way described below is not tied to a specific model and will work for 1050, 1070 as well as other models. In the program window, the stock video card configuration looks as follows:
The first thing to consider is overclocking the Nvidia GPU for coins on the Ethash algorithm, which include: Eth, Etc, Ella, Pirl, etc. Their complete list can be found on the same Whattomine or cryptocurrency listing. To mine these coins, you need to overclock the memory and leave the core at zero. So, we take the slider called Memory Clock and start dragging it to the right until it reaches 300 MHz. Then slow it down and add 50 MHz, watching how the card behaves.
As long as the GPU is running stably, continue integration, as soon as any bugs appear, roll back the slider to the previous value. In general, most cards can easily handle 500 MHz, but it is better to play with the approximation.
Now let's work with power consumption. It is necessary to achieve an optimal ratio of power consumption and stability. To do this, we set the Power Limit. Its level depends directly on the card type. For example, for our card 1060 6gb it is 65% and for 1050 ti it is 70-75. Here again, we have to play with approximations. We move the slider to the left until the hash rate starts to sag, and then return to the closest value at which the performance was kept at a stable level.
At the end of the day, we got the following results:
Memory Clock - 600.
Power Limit - 65.
With these settings, the card produces about 23 MH/s instead of the original 19 MH/s. So we see that Nvidia Inno3d 1060 6gb GPU gain more than 20%.
- OVERCLOCKING NVIDIA GPU FOR ZCASH MINING
For coins based on the Equihash algorithm, such as ZCash, Hush, and others, overclocking of Nvidia GPU is done by a slightly different principle. In this case, we will have to work with the frequency of the core and the power consumption limit. So bravely take the slider Core Clock and pull it to the value of 100, gradually increasing by +10 to reach the limit value, at which the card works fine.
Then we go back to Power Limit again, pulling the slider to the left. The main goal is to achieve a high hash rate with low power consumption. What we have got is the following:
Power Limit is 85.
Core Clock is 140.
One more thing to note about the Fan Speed indicator - there's no point in regulating cooler speed in wintertime. Even the stock settings are fine, especially if the room is not heated. In summer, all this will need to be adjusted with handles, as the automatic adjustment is not always effective.