How to debug your mining rig

Your mining rig won't start, doesn't detect all its graphics cards or crashes after a few hours? In this 2022 video (in French), we go through the most common faults and how to diagnose them, first on a new rig that has just been built, then on a rig that was already running and starts causing trouble. Most of the time, the culprit is not the graphics card but a riser, a cable or a setting.

New rig: check the basics of the build

There is not much to install on a motherboard, but nothing can be forgotten:

• The CPU cooler must be fitted;
• The RAM: even a DDR4 stick that is compatible on paper (2133 or 2400 MHz) can cause problems. Reseat it, try another slot or another stick;
• The power cables for the motherboard (24-pin) and the CPU must be connected.

If the rig doesn't seem to start, or makes a small "click" at power-on and goes no further, something is wrong: either the power supply is dead or a cable is badly connected.

Not all graphics cards are detected

Don't plug everything in at once: add the cards one or two at a time to find the faulty riser or cable, switching off the power before each change. To see which cards are recognised, set the fan speed to 100%: only the detected cards spin at full speed. Riser problems are no myth: replace the riser and check the connections before suspecting the card.

If the connections are fine, move on to the BIOS:

• Above 4G Decoding must be enabled (it is by default on some motherboards);
• Boot mode: there are two, CSM and UEFI. If cards are not detected, try disabling the Launch CSM option;
• PCI Express speed: set the PCIe ports to Gen1.

A real example: on a mining motherboard with eight PCI Express ports, seven cards on the small ports worked fine, but with the eighth on the main port, HiveOS only recognised seven out of eight. Disabling CSM was not enough. What solved it: setting every port to Gen1, including the main port, in the BIOS PCIe configuration menus.

The rig crashes after a while

A rig that mines normally, then crashes and reboots on its own, is often suffering from an overclock that is too aggressive, especially on memory frequency, or from overheating: it's summer, it's hot, the cards' temperature creeps up without anyone noticing, and some models such as the RTX 3080 are particularly sensitive to it. To find out:

• lower the memory overclock;
• raise the fans to 100%;
• check that the rig stays stable for several hours before going back to your settings, using the overclocking examples published online for each card model.

A card disappears or shows up incorrectly in HiveOS

If a card that was working is no longer recognised while mining, the riser is suspect number one: replace it and check that no cable has worked loose.

Another symptom: HiveOS doesn't show the model name (for example "GeForce RTX 3070") but the chip's code name. This means the card was not loaded correctly. Again, replace the riser and check the card's power cables.

The rig powers on but HiveOS doesn't start

When the rig no longer shows up in HiveOS, you need to isolate the faulty part:

• The power supply: inexpensive PSU testers are available. You plug the 24-pin connector into one and it tells you whether the power supply works;
• The motherboard and RAM: take the motherboard out of the rig to test it on its own, and check the memory stick. A spare motherboard (about €150 at the time of the video) is a useful investment to diagnose quickly and replace a faulty board;
• The HiveOS drive: if the system is installed on a USB stick, its file system can become corrupted and HiveOS no longer loads properly. The problem can also come from a faulty installation on the USB stick or the SSD. Reinstall HiveOS or switch to another drive.

What about today?

When the video was made, Ethereum's move to proof of stake had just been postponed to September or October 2022. It took place on 15 September 2022 and ended Ethereum mining with graphics cards. The troubleshooting method, however, still applies to any GPU rig: test the parts one by one, and check the power supply, the cables and the BIOS settings.

To go further, read our article on motherboard installation and BIOS setup for mining, our 10 tips to keep your rig cool in summer and all our tutorials. Stuck with a fault you can't fix? Contact us.