High Temperature: Efficiency Killer

High temperatures can significantly impact the lifespan and performance of ASIC miners. Excessive heat can lead to reduced efficiency, and, in severe cases, permanent hardware damage.

ASIC miners are composed of numerous transistors and other semiconductor components. As these components heat up, their resistance increases. This means that to push the same amount of current through the now-higher resistance, more voltage (and thus more power) is required. The increased power required to overcome the higher resistance is not used for productive work (i.e., hashing), but rather it is dissipated as heat. Therefore, the overall efficiency of the miner, measured in J/TH, decreases.

Using a brand-new Bitmain T21 as example, we can see how performance is significantly influenced by changes in ambient temperature. As the temperature rises throughout the day, the miner’s power consumption increases markedly. During peak afternoon temperatures, this increased power demand leads to reduced efficiency, impacting the overall performance and profitability of the mining operation.

Conversely, as ambient temperatures decrease in the evening and night, the Antminer T21’s power consumption drops, highlighting improved efficiency. Cooler conditions reduce the electrical resistance, enabling the miner to operate with less power while maintaining optimal performance. This pattern underscores the importance of effective thermal management strategies. By ensuring a stable and cooler operating environment, miners can enhance the efficiency and longevity of their equipment.

From the lowest to the highest power consumption observed in the provided graph, there is a 250W difference, which translates to a 7.3% variance. Operating at its nameplate settings of 180TH/s, the efficiency of the Antminer T21 fluctuated between 19 and 20.5 joules per terahash (J/TH).

Cold Temperature: Curtailment Enemy

The thermal stress caused by wild temperature swings can result in thermal expansion and contraction of the hashboards. This constant expansion and contraction can lead to physical damage, such as microfractures in the board, and can cause components like heat sinks to unsolder and detach. When miners curtail in cold environments, the machines stop producing heat and can cool down rapidly. The thermal shock caused by the swift temperature drop can cause significant stress on the components, leading to potential damage from rapid contraction of the material.

A solution to avoid the thermal shock is to slowly allow a progressive cooldown of the hardware. By gradually underclocking the miners over a few minutes instead of instantly cutting power from the miners, miners can ensure a smoother transition to lower temperatures.

Here’s an example to implement a smooth curtail using LuxOS Firmware and LuxOS Tooling Python Package. The script will underclock a list of miners one profile at a time every 60 seconds. Once it reaches to the lowest profile it will issue a curtail command.

import time
import asyncio

from luxos import utils

if __name__ == "__main__":

    addresses = utils.load_ips_from_csv("miners.csv")

    # LuxOS comes with 16 built-in underclocking profiles
    # Underclock one profile at a time every 60 seconds
    for item in list(range(1, 17)):
        asyncio.run(utils.launch(addresses, lambda h, p: utils.rexec(h, p, "profileset", parameters=f'{item*-1}'), batch=None))
        time.sleep(60)

    # Finally curtail
    asyncio.run(utils.launch(addresses, lambda h, p: utils.rexec(h, p, "curtail"), batch=None))