Bitcoin Forum

Mods, please, oh please do not merge this with the other thread. People need to see the subject line, as this has a day's worth of data (work) on my part :)

Firstly! I hope Bitmain continues to release firmware with this kind of adjustability. Also note that this F/W disables SSH, so if that's important to you don't use it.

This is really nice firmware, it gives you a total of 13 settings counting normal.

https://i.imgur.com/dgQlpZe.jpg (https://i.imgur.com/dgQlpZe.jpg)

Below are the preliminary test results, I say preliminary because each setting was run for roughly 30 min. Ideally each one would be run for 24 hrs or so, but I did not want to wait 2 weeks. Power usage and ambient temps were taken from an APC AP8941 "smart" PDU. Other data taken from the miner status page. Run time ~ 30 min each which is three polling cycles on the PDU. The S9 used to test is a February 2018 13.5 Th/S S9 with an APW3++ PSU. Please note the ambient temperature as this room is not temperature controlled and varies as the day heats and cools.

https://i.imgur.com/3rpAvgj.png (https://i.imgur.com/3rpAvgj.png)

Here's some rough earnings estimates at current difficulty and $0.06 kw/hr pricing, it's sorted by daily net income.

https://i.imgur.com/nmU36ls.jpg (https://i.imgur.com/nmU36ls.jpg)

Is this the S9-xilinx-201905311059 version?

Great data, now I need to run the numbers and figure out what SSH port it/does and if I need that ;)

Um missed the part where he says


?

Any idea what the voltage reads?

Good question, I did not test, but it would be good to see the frequency and the voltage of each F/W settings. What is the best way to test that? Can it be queried from the API? Read only is still allowed.

Would you say the older LPM version is better overall?
Seems it runs a good bit cooler, small watt savings, and not much of a hit on the TH as well.

I'm not sure I would say better, I would say LPM is a nice all around f/w and pretty much required if you need/want SSH access.

I ended up flashing 2x S9 and 1x S9i to the new firmware and running them all at -0.5TH. After 12+ hours I am showing more than a 60W decrease, so far it's actually close to a 100w savings with a very similar ambient temp as yesterday. Maybe this is due to autotune settling in and saving wattage on fan speeds. Not sure, I definitely need to let them run much longer for a good average and I'll be able to report back with better accuracy.

my s9 13.5 with brains runs 15th/s just under 1300W Feq 675 voltage at 8.9

I believe you'd have to measure voltage off each board.

Also, my 14th/s on older Bitmain LPM FW runs 14th/s at 1230W. Will have to do some math but potentially 16th/s 1430 isn't too bad.

The T9+ also has the new firmware.

Original info from my other post about it that is now locked:
So far (3 hours) running with the "hash rate unchanged- lower voltage option"
I have seen a SMALL rate drop from 10.3T or 10.4T down to 10.1T or 10.2T
I have seen the input power drop from close to 1500 to 1230

As of now 19 hours in.
Power is still down BUT hashrate has never gotten about 10.125 to 10.14

Also there is no low power enhanced mode that the S9 has.

-Dave

Exactly why would they disable ssh for any reason? Like you can just lock us out of our own miners...

My guesses would be below in no particular order.

1. Less support tickets because people stop blowing up the software
2. Less support tickets by locking you in to specific settings so people stop blowing up the hardware (faster)
3. Stopping people from installing 3rd party firmware

braiins may be better but this is decent if you don't want to use aftermarket firmware.

I am playing with an s-9 space heater.

So this or braiins may be just what I want.

I figured a way to keep miner quiet. Using other fans and only 1 board.

Here is my gear with

 LPM setting
 1 board  mining
 hash is 3300
 watts is 350

Wow, that's really good. What is the DB reading? I've been down-clocking an S7 just via it's web interface, but even at 3.6 Th/S (575 freq) it still uses 1000W. It's out with my other miners so I have not been able to measure the DB.

3 is the likely reason. Braiins OS single handedly extended the life of gear they had planned to retire earlier, forcing them to come up with nonsense like the s9e which comes with a similar firmware.

Regretfully the T9+ is not working with Braiins OS yet, and this Bitmain firmware doesn't let you control speeds and voltages like Braiins OS does. Experience with Braiins OS using S9s has demonstrated that no asic hash board is equal, and global settings like those offered by Bitmain are subpar at best.

Also with full control you can make your own LPM modes, more reliably than a naive on/off switch that causes trouble to some owners (especially with older hash boards).

As wonderful as it sounds it's still 3rd party software. Which will discourage some miners and pool operators. I am not personally knocking it, just stating that.

Well this is one of those few instances where the third party is superior to the "original" for various reasons, starting with full free open source, which is about the same reason you would run Linux instead of Windows in your computer.

Miners can do whatever they please with their gear, or at least thats how it should be. Some companies think they should keep controlling the things they "sell", treating them more like some sort of lease you have no rights to...

As for pool operators, its their choice to accept or reject hashrate they could consider "harmful" for whatever reason. As a miner, you are also free to choose a pool where you are welcome.

Do note that Braiins OS does NOT require you to use Slushpool, and it doesn't even start mining there by default (tho its shown in the default config, Unlike Bitmain's, it actually doesn't start hashing straight away to their address...

S9 code is open source.

Meanwhile the other 'so called' better firmware added #xnsub to their code (which probably is in some versions the S9 code also - no idea)
2 things about that:
1) It's a major security issue since it allows the proxy/pool to change the work being mined to without any notification or log of it happening.
i.e. the pool/proxy can even switch what coin or what pool or what username you are mining without any log of that.
2) Who knows what issues they caused adding the #xnsub code ... did they test it still finds blocks?
(no doubt the standard reply to that is "Why do you need to find blocks?")

This relates to destroying hardware and sending it back to bitmain wanting them to fix it ...

While it's under warranty you actually can't do whatever you want to it and expect them to fix it ... which no doubt is the problem there.

Bitmain should add some hardware log of max temperature/frequency/voltage that can't be deleted - but no doubt someone will come up with a firmware to hide that.

People like to make up conspiracy theories about all sorts of rubbish, but it seems pretty obvious why they would do it if you take a very small moment to think about it logically.

I tried the +2th with no issues, but just worried about burning something up.

Anybody else running 15.5th on the 13.5 S9? Is it stable? Any power draw numbers?

So what is the equivalent of LPM mode for this firmware? Is it the 'Hashrate Unchanged Lower Voltage' setting?

I would say it's best to test as each miner is a bit different, but the "Unchanged Lower Voltage" was close as was the -1 Th/S setting. One just above and one just below.

I am getting 15.4 but using only the +1.5th , the +2th does not seem to be stable for me, just as OP mentioned, I also have no explaination on why does +1.5 adds 1.9th but well I can't complain.

also keep watching the rejected shares, or basically watch the reported hashrate on the pool, don't get too excited seeing the hashrate on your miner , after all it kind of does not matter if it's not reporting the same on the pool due to excessive over clocking, this happened to some of my S9s even with +1.5 so had to go back to +1 on those.

Sure justify draconian corporate control over hardware, why Tivo did it and some computer manufacturers (secure boot) and smartphone manufacturers, doesn't mean its moral or that people have to accept it, otherwise lets forget open source and go closed source, and while we are at it, lets dump this choose your own pool rubbish and go Bitfury, see how you like that.

Warranty is no excuse, touch the firmware and lose the warranty. I'll let your imagination figure how easy it is to detect that. But I'll let you a tip: Bitmain is incapable or unwilling of detecting subtle hardware manipulation, as in hash boards changing places... Its probably a logistics nightmare and cheaper or better to just ship new units rather than lose a customer. Something you seem to ignore here: Bitmain is that profitable in their core business. The minuscule amount of burned boards due to customer abuse means nothing to them. Oh, and they can get burnt in a myriad of ways without users messing with voltages and chip speeds...

You are pissed over #xnsub when the actual Bitmain firmware supports it. I'm pissed when people install Windows and yet they still do it knowing the risks. such is life. Leave it in your sig and call it a day. People still go to nicehash despite having taken customers money... #xnsub is optional and unnecessary for most users anyway, I'm willing to bet most people learned about its existence thanks to you, talk about Streisand (https://en.wikipedia.org/wiki/Streisand_effect)...

I'm glad someone wrote open source code to replace Bitmain's, and its been a success, despite supporting only S9s and T1s. No hardware should ever be tied to an specific piece of software, ever.

Tell you what: go github, fork their project, call it safebraiins or something and remove the #xnsub code. Anyone could do it, that's the beauty of free open source. You could even remove the "dangerous" overclocking options...

Aren't you a developer yourself? You really should know better: code talks.

Also you neglect under-clocking and energy efficiency advantages. While some users (foolishly) go the other way, this firmware lets you run them quieter and more efficiently, thus prolonging their lifetime. Bitmain is unable to carefully adjust this to every particular hash board for whatever reason, but users can and are willing to find the best speed that produces the most hash while consuming the less amount of watts per hash, per board.

And, there are boards that show dead using Bitmain's but work with this, that are long past their warranty. Ever heard of e-waste? This is the same as older computers getting tossed because Windows 10 stutters with them but work fine with Linux. Again, prolonging lifetime beyond its expected life-cycle by replacing the bundled software with a free open source alternative.

As for "conspiracies", well, there was Antbleed (https://cointelegraph.com/news/bitmain-claims-antbleed-had-no-malicious-intent-does-it-really-matter-for-bitcoin-miners) but i think have already given enough reasons...

Hah - it didn't do anything - check the code.

Conspiracies indeed :D

I even posted about it well before that.
However, it didn't allow them to stop the miner.

Fun how people believe anything they read and wont verify if it's false when they want it to be true :)

Is it possible to have a batch that just does not like +ths at all?

As soon as I use any of the options to add TH/s my miner, s9j, loses TH and goes around 12th instead of at least 14,5.

Now the interface shows the correct speed, it includes the additional TH, but I checked on the pool and there I can see it is lower.