Antminer L3+ Overclocking – Is It Worth the Squeeze?

antminer-l3+-overclocking

Like many of you, I frequently wondered how much more I could possibly squeeze out of my Antminer L3+ / L3++ before smoking it completely. I conducted research to determine the extent of the harm that can be caused at various overclocked frequencies for my own education.

Disclaimer – These are merely some baseline levels to take into account because every L3+/++, hash board, and environment in which we run these are unique.

Antminer L3+ Standard Factory Firmware and Base Settings.

Let’s continue using the base Antminer L3+ with its default factory firmware and default settings. Using V1.5 hash boards, that gives us an operating frequency of 384MHz and a voltage of 10.11VDC per hash board IC.) So, we now have a baseline of 504 MH/s (126 MH/s for each hash board).)

We can get 740W total (185W per hash board) by subtracting about 60W for fans and the control board from Bitmain’s stated specifications for the base L3+, which state an 800W nominal operating power.) Even so, let’s just reduce that figure to 200W per hash board because, according to my research, they don’t specify whether they will include fans or a control board.

Bitmain Antminer L3+ Specification

Quick calculations give us an estimate of 1.59 W/MH, which is nearly exact to Bitmain’s specification of 1.6 W/MH. I operated one of my L3+ units at this setting for a few hours, and as can be seen in the chart below, the Temp(Chip) of each hash board remained at a very reasonable level. However, what’s interesting to note is the fluctuating W/MH and Temp(Chip) levels over the course of the experiment.

Some observations about the testing: It takes roughly that long for the temperature and MH/s rate to stabilize, so I ran each speed/voltage for between 10 and 15 minutes. The room’s average temperature was 72 degrees, and the fans rotated at a variety of speeds, most frequently between 5000 and 5400 RPM. Ohm’s law shows that each hash board needs about 17 to 20 A of 12 VDC, which is another quick learning point for me.

Three Test Voltage Settings of Antminer L3+

The following three graphs display the three test voltage settings (9.5V, 9.8V, and 9.92V) as well as the relationship between frequency (x axis) and MH/s(AVG), as well as W/MH (y axis).)

The graphs show that the actual (MH/s(AVG)) is actual normalized rate, which essentially means that this is the number that is closest to the actual work you are performing. Since MH/s(RT) is an instantaneous value that ignores Nonce%, it may give you a misleading impression of how much work you are actually doing. What does that mean? As you can see in the following graphs, the AVG rate and RT rate remain consistent up until a point where we begin to experience significant HW errors, which have a negative impact on our Nonce%. For those who are unfamiliar, the Nonce% is the proportion of HW errors to Nonce (numbers used only once).) Both the Nonce% and the DiffA% must be reduced. Generally speaking, if you’re not severely overclocking and under-volting, you’ll see a low number of hardware errors that will keep your Nonce% under 0.03% (I’ve heard this arbitrary number many times, but I still don’t understand it).) Since the Nonce and the DiffA (difficulty of last accepted share) are somewhat correlated, keeping the Nonce below 0.03% keeps your DiffA at or below 0.0002%.

AntMiner L3+

500MHz Setting on Antminer L3+

In conclusion, if you hear about some incredible overclock settings where someone is getting 650-700 MH/s from a Bitmain Antminer L3+, that’s probably their RT and not the AVG, or actual work. Although the data shows that the actual AVG rate was closer to 575 MH/s than 650 MH/s(RT), I ran up to a 500MHz setting and achieved that.

Thanks for the information, it’s all good. Therefore, what does that mean for me and what are the best settings? For the simple reason that every board and circumstance is unique, neither I nor anyone else online can provide you with an answer. Through this information, I hoped to reiterate some general best practices. Before temperatures that could harm your boards (typically under 75–80C) are reached, there is a point of diminishing returns.) Below are a few graphs illustrating the varying temperatures and the slight increase in operating frequency.

Conclusion: Antminer L3+ Overclocking

There are many different firmware configurations that can autotune; I personally use HiveOS on most of mine, and they do a good job of fine tuning. However, you must choose the operating frequency before they can tune your L3+ in. The data suggests a sweet spot at about 450MHz at 9.92V if you are in a setting similar to what I described. That serves as my fundamental starting point.

Scroll to Top