Feature request:

Add a flag / parameter to inhibit Auto Tune, thus using / keeping the current tuning results.

Why: The rig had been running fine. Suddenly, reason, unknown, auto-tune kicks in for one or more cards and it spends the next 12 hours doing it's thing. Sometimes the results are even worse than before. So 12 hours are hashed at 1/2 the normal rate while tuning.

If I can catch it, I will reload a saved configuration. That takes jut a few minutes to get back to the normal hash rate. If not (or don't have a saved config.), and the results are worse, I will then have to set the  power rate at a watt or two higher than before to get auto-tune to kick in (or edit the auto-tune file), but that's even worse as now the whole rig is tuned.

The software saves the best frequency for each chip at a certain card voltage in a unique json section for each card, for each power level. I'd suggest using that data when starting a tune, vs blindly cyclying thru all the combinations.

Oh, and make auto-tune enable & power level selectible each card. I may want to set a given voltage / freq. for a particular card while other cards auto-tune at each specific power level.

This happened to me too.

The problem is that at startup / warmup the software sets the fan speed to a very low value (1% in my case)  and then it complains that the fan is missing when it sees the rpm value drop below a certain (unknown) level. My fan was doing 120 rpm with the 1% command.

I opened a support ticket and was basically told it was all my problem.  That they had seen this with 3rd party fans.

I offered the solution of not checking fan rpm minimum unless the commanded speed level was above, say, 10%. This type of checking is normal for anyone that has done actual real world hardware control.

They said that they may include a "feature" sometime in the future that will allow a user to set some minimum level.

Have you fixed the problem with the S9 version that does not shut down the miner when internet connectivity (or all pools down) is lost? My rigs just mined nothing - producing heat - for over two hours today when the power supply for a network switch went belly up.

As they are in low power mode and sound proofed enough they are 10 ft away. I can hear them, but not be distracted. For the whole 2 hours the fans kept going and when I checked hot air was coming out. After replacing the power supply the network connection resumed. I waited almost 10 minutes and they still had not reconnected to a pool. So I had to reboot them. This is not good at all. They have static addresses.

Another thing to fix: When enabling / editing pools it wants to do a warm start and do the tuner testing hash rate routine, this using 5-10 minutes to do so. Unless the power setting is changed I can't think of any reason for this.

And lastly. Please make it so I can save / restore configurations to my computer. Just like the original Bitmain software.

Thank You.
 

I am using version "LEDE Reboot 17.01.4 r3560-79f57e422d / LuCI bos+/stable-21.04 branch (git-21.099.47497-5c3e5f9)" on an SD card for a couple of S9's.

The fans constantly hunt up and down at least +/- 80 rpm, changing speed about every 6 or so seconds. Chip temperature is 80C and varies +/- 0.15C. Although the S9's are in a location with noise abatement, it's the sounds of the varying fan speed that's annoying. If the speed was more constant (which happens with other S9 software) the sound would be much more tolerable.

Seems to me that there should be more dead band in the control loop. We know power draw and therefore heat varies constantly as hash rates change every second. So trying to chase the temperature with varying fan speed is pretty fruitless.

Are the temperature loop tuning parameters exposed in any way so that I can adjust them?

Why isn't he just letting it auto-tune? My experience with AM firmware (and all the derivatives of the main source of it) on an S9 is that auto-tune can do a decent job and then some hand tweaking after that can be done.

But what I have seen is that after auto-tune and hand tweak, it'll show great for many hours / days and then things will turn downhill. For what reason I'm not sure but I did have to go back in and do some more hand tuning. 

Just for comparison my 2 S9's  are running an average clock of around 713Mhz avg., 8.4x volts for 15.25 TH/S but that is with asicseer firmware, 1250-1300 watts, 240vac, ambient ambient 60 F, S9 temp. at 70 C with fans at 2400rpm.

That's no better than factory stock firmware. Are you mining with AsicBoost ? That's the only thing I would think that would be the cause. Some pools have different URL's for AB. With AB On, you would, with stock Bitmain firmware, be drawing only around 1250 watts, at the normal 13.5 hash rate. With AB on and 1440 watts you should be at least 15 Th/S, with factory firmware. I know with other 3rd party  firmware I got a tad over 17 Th/s at 1550 watts

I had thought that I read somewhere that BOS could adjust the frequency on every chip. Unfortunately, after reading all 32 (?) pages here and searching, I cannot find that again. Can someone enlighten me on where this is in the web interface. I've installed it, can only find where I can set just the voltage / frequency per board.

 And (this may sound snooty - but it's not meant to be) any other actual documentation for the using (not installing) of BOS? Especially from a not-so-much-unix persons view for parameter files and command line directives.

So I'm new to this.

Using a projected difficulty level and after the having of BTC in the spring of next year, will an S9 be obsolete? Like so many other miners on eBay, not profitable to mine, can't even give away, considering electricity cost? Even at my $0.0395 / Kwh cost my calculations (with projected difficulty) after the halving indicate negative return.

Yes / No / What am I missing here? Enlighten me please.

Not quite. You have 220/240 in your home for the electric stove,  clothes dryer, A/C unit & ....

While there is a step down transformer on the pole (or in a box) for distribution lines at 13,800 volts (typ.) down to 240 secondary side, there is a center tap to get 120 from each leg to the center tap for the 120v  "neutral".