All warranties of AMD stuff need to go through the manufacturer. I doubt that the retail shop that sold it would replace it. They could see that you have a modded bios but in my experience they will RMA the card.

You will never make your money back running one of these. As it sits the ROI is nearly 2 years at current prices, meaning you will be priced out by the difficulty adjustments and never break even. Then you add in the costs of a powered hub and you are just throwing money away. I just dont get it....

After the difficulty bomb last week ZEC has been a good 30% more profitable than ETH to mine.

This only matters if you are immediately selling the coins. For some reason people think that if they mine the coin while its most profitable they make more. This is absolutely not the case is the only number that matters is the price when you convert it to something else.

Then, of course, there are the people that just throw their money away using nicehash because they are too lazy to convert it for themselves.

Because different algorithms process at different rates. You absolutely cannot compare the mining speed of one coin to another, it simply doesnt work that way.

All the people saying the CPU could be to blame are just plain wrong and dont understand how PC architecture works. Even on my 12 card rigs the G3930 Celeron is only at 25-30% usage at max. Could you be a bit more concise in your problem description? Cards 'acting up' could be anything. If you give some specifics it would help the troubleshooting process quite a bit.

Your miners will vary in temperature based on the ambient intake temperature. What exactly are you asking?

You need to give at least some details, maybe a screenshot of your status page. We cant read your mind and pull the details out for ourselves.

What edge are they supposed to have? They just have a facility in Washington just like a bunch of other people. They keep trumpeting how big they are but their press releases sure dont make it sound that impressive. Claiming to be an industry leader with only 18ph of machines is a joke.

This is so far off base I laughed out loud when reading it. Noone pointed it out because you are 100% wrong in everything you stated. Ebang was the contract manufacturer for the Avalon A6 so if there is any comparison to be made its that they cloned the Avalon. How do you miss that in even the most basic comparison? All you have to do is look at the design. Then you claim they are lying about using their own chip. If they used bitmain chips dont you think the efficiency would be more in line with the S9?

Avalon = 2 boards per unit = 44 chips per board
Ebit E9 = 2 boards per unit = 48 chips per board
Antminer = 3 boards per unit = 63 chips per board

Avalon = 0.16 j/gh
Ebit E9 = 0.15 j/gh
Antminer = 0.10 j/gh

What effect does the HBCC have on mining? From everything I read it is a technology with no real use. In lab tests they had to absolutely destroy the card running 8k resolution benchmarks to even see HBCC in use. In real world gaming and workstation applications it is 100% useless.

This newer riser just gives you more options to power them. You only need any one of those 3 connectors to be powered for it to function properly.

Seems like a really poor design having the molex and the PCIe pointing up towards the card like that. It would be a real mess powering them that way.

Sadly this is not said/explained nearly enough. People keep chasing what is most profitable for the day or week but if you dont immediately turn your coin into fiat there is literally zero reason to chase daily profits. As the quoted post says, the profit comes from hodling your coins. Mining is a long game, despite what the majority of people seem to think.

There is no need for any memory past 8gb on cards today. ETH will go proof of stake before the DAG gets that big. What you are modifying on a card is memory and processor speeds/timings and voltage to try and get the card to its maximum speed with a minimum of power used. There is a lot to be said for these kinds of mods as you gain both power efficiency and hash rate which in the long run will earn you much more profit.


Also, you dont want to try and find knock off versions of these cards. Your performance suffers greatly even on a true retail card with the wrong brand of memory. You need all the components to be quality if you want the card to last and make you a profit.

I wonder why it says it only supports Windows. Thats why they claim it only supports 16 graphics cards but is there any reason why linux wouldnt work on this? Its still just a B250 motherboard. Strange they would add 19 slots then tell you that you cant use them all.

You also have to remember those older miners were using much older manufacturing technology. As the dies shrink on the ASICs the problems increase. You cannot compare any miner made in the last 2 years with anything older, its like apples to oranges.

Being conservative I would say a full speed s7 pulls about 450w per board + 50w for the controller which powers the beagle and the fan.


If you are running 2 PSUs I would run them like this


PSU 1 = Hash board 1 + hash board 2 = 6 PCIe plugs required

PSU 1 = Hashboard 3 + Controller = 4 PCIe plugs required.

Make sure you turn on PSU 1 first so that when PSU 2 boots the controller, it sees all of the hashing boards.


NEVER connect more than one PSU to a single board. This will fry your hardware.


As long as PSU 1 is rated for 1100 watts or more you should be fine.

As long as PSU 2 is rated for 600 watts or more you should be fine.

I have managed a fair number of A741s and the overall failure rate after 60 days including DOA is under 2%. This is far and away the best I have seen from an asic hardware company.

With such a tiny farm there really is little need for monitoring software. There are many pools out there you can set to send you downed/slow machine alerts via email.

As far as mass pool changes there is actually a bitmain tool linked in another thread that will do whatever you need. You can use it to pull live stats from all the machines as well as re-pooling and a few misc changes I did not list.

Mining solo is a terrible idea if you want regular profits. Using nicehash is just as bad since you are basically giving them a portion of your revenue for no reason.

You can literally run 50 miners off of an ISDN connection, there is no need to run an internal proxy or anything else like that with your setup.

I wouldnt recommend any of the mining software you mentioned (minera, cryptoglance, awesome miner) because they are either made for a home user and do not scale very well when you run a fair number of miners. Others just cost too much and are inflexible with their licensing at the large scale.

If you dont vent the area you are mining in, things will overheat within a few hours of turn on. You NEED active airflow to keep things cool.

L3 = 900w space heater
D3 = 1200w space heater
S9 = 1500w space heater

You have to look at the units like that and plan accordingly.

So ill have to disagree with the claim you should never use multiple PSUs to power a miner. As long as they are connected correctly this is not an issue. I ran several thousand Avalon6 units with dual PSUs with no issue.

The second statement should be read and re-read if you are going to try this. Never have more than one PSU running an individual hash board, otherwise they will go up in smoke quite quickly.

Also with the way the Avalons work if one PSU is failing you should still be able to mine with the other board still. It should be pretty easy to tell which PSU is causing the problem.

Last bit of advice, you should not be running ATX PSUs on industrial mining equipment. ATX PSUs are not meant for 24/7 usage at near max capacity. You should be using a server PSU of some sort, they were built to be leaps and bounds more reliable and efficient versus an off the shelf ATX PSU.