Stratum v2: After 10 Years, The Most Used Bitcoin Mining Software Gets Facelift

[FP91G]

Mark the Stratum v2 release as a victory for the open-source bitcoin community. For a decade, it has been the software of choice for miners to interact with pools and with the bitcoin protocol per se. While still necessary, bitcoin mining pools have a centralizing effect. With Stratum v2, miners will get to construct their own blocks. And that’s just one of the innovations, although the most talked-about one.

https://bitcoinist.com/stratum-v2-bitcoin-mining-software-gets-facelift/
https://stratumprotocol.org/

[NotFuzzyWarm]

V2 has little to do with the original Stratum.
A miner should do only three things : get shares from a pool/node, process the work, report results.
Oh, and monitor itself & provide a GUI. Fine, that's 5 things. Period, end of story.

IMHO having the low power CPU's in a miner deal with more than the absolute minimum of work is a horrible idea. Then there is the matter of changing work w/o it being obvious to the owner of the hardware. Ya know - like the security hole known as extra-nonce aka XNSUB which allows the miner to change work on-the-fly to mine other coins (DevFee hashing is 1 example) -- as well as do other things in the background if it connected to a bad-actor pool.

Everything I've seen points to V2 being solely for the benefit of pool operators and not the miners themselves.

[DaveF]

Lets see if it gets any traction in the real world outside of big corporate miners.

I can see it slowly taking over as the manufactures start using it by default, but until you hit that critical mass point I see most people still using V1, I know a lot of people never even update the firmware on their miners unless thy wind up having an issue with something. For the small miners, there really is no advantage to all the additional overhead that it needs. And as with anything, the more features you add to something the more issues you will probably wind up with.

-Dave

[_oh_no_stop_this_]

the more features you add to something the more issues you will probably wind up with.

That is so true

Features of V2 seem to be cool, but there are so many things... is it all expected to be configurable from within each individual miner? Or is it expected that we'll have a LAN device that aggregates traffic and provides configuration interface?

Upd: I see they have different connection options on their website. Is there a way to monetize this so-called "translation proxy" that is basically what I mentioned above? Maybe there is a commercial opportunity, but still I don't know why would anyone buy such device.. there is no real financial incentive.

[kano]

Firstly - block constructing was called GBT. Yeah years old idea.
Some pools supported GBT for a while ...
No one ever actually wrote code to use the transactions supplied - GBT died.
It's death was a good thing.

Stratum already has the option to request the transactions, so you can verify the merkle slices, but not change the ones used.
No one uses GBT or that option in stratum for a reason:

What it actually leads to is 2 things:

1) about 2000 times more data per works sent, and thus about 2000 times more data per day per miner.
Yeah so instead of about 15MB a day per miner per pool connection, you are looking at about 30GB per day per miner per pool connection.
This also means on average more than 3MB of data per work change instead of less than 1500 bytes.
Networks do not transfer data instantly.

2) transaction bias
This is completely against the ideals of bitcoin.
People have tried, on multiple occasions, to put this into bitcoin. Fortunately they've failed doing so.
Adding it to pooled mining is a really bad idea.

----

The initial problem with stratum was the fact that the work difficulty was separate from the work.
Slush wouldn't fix this coz it meant changing his pool code. Very slack of him.
However, not a big problem, unless a pool is misbehaving or coded badly ... ... ... ... ...

[Daltonik]

Mark the Stratum v2 release as a victory for the open-source bitcoin community. For a decade, it has been the software of choice for miners to interact with pools and with the bitcoin protocol per se. While still necessary, bitcoin mining pools have a centralizing effect. With Stratum v2, miners will get to construct their own blocks. And that’s just one of the innovations, although the most talked-about one.

https://bitcoinist.com/stratum-v2-bitcoin-mining-software-gets-facelift/
https://stratumprotocol.org/

What we see on the site stratumprotocol.org on its main page it says this: Stratum V2 is the next generation protocol for unified mining. This increases security, increases the efficiency of data transmission and reduces the requirements for the mining infrastructure.
There is no question of any individual creation of blocks, the protocol simply improves the interaction between miners, the pool and the blockchain to increase the efficiency of joint mining.

[mikeywith]

What we see on the site stratumprotocol.org on its main page it says this: Stratum V2 is the next generation protocol for unified mining. This increases security, increases the efficiency of data transmission and reduces the requirements for the mining infrastructure.
There is no question of any individual creation of blocks, the protocol simply improves the interaction between miners, the pool and the blockchain to increase the efficiency of joint mining.

The blocktemplate creation by miners is an add-on feature, it's not the backbone of Stratum V2, the miner firmware needs to implement it and the miner owner needs to choose whether to use it or not, many people are skeptical about this, some people think things could wrong and pools would start losing blocks due to miners adding unverified transactions by mistake or for any other reason.

Personally, I think it's too early to judge, I believe the extra encryption and using plain binary instead of JSON will lower the bandwidth usage which is great (of course, assuming the miner doesn't care about the transactions and just sticks to the V1 principles of receiving work and sending shares to the pool).

Such a transition would probably take years, but I am confident that sometime in the future every pool and every miner would be using Stratum V2 after it has matured and become stable.

[DaveF]

What we see on the site stratumprotocol.org on its main page it says this: Stratum V2 is the next generation protocol for unified mining. This increases security, increases the efficiency of data transmission and reduces the requirements for the mining infrastructure.
There is no question of any individual creation of blocks, the protocol simply improves the interaction between miners, the pool and the blockchain to increase the efficiency of joint mining.

The blocktemplate creation by miners is an add-on feature, it's not the backbone of Stratum V2, the miner firmware needs to implement it and the miner owner needs to choose whether to use it or not, many people are skeptical about this, some people think things could wrong and pools would start losing blocks due to miners adding unverified transactions by mistake or for any other reason.

Personally, I think it's too early to judge, I believe the extra encryption and using plain binary instead of JSON will lower the bandwidth usage which is great (of course, assuming the miner doesn't care about the transactions and just sticks to the V1 principles of receiving work and sending shares to the pool).

Such a transition would probably take years, but I am confident that sometime in the future every pool and every miner would be using Stratum V2 after it has matured and become stable.

Been thinking about this since you posted it a week ago. I keep looping around. Are miner manufacturers going to create new binaries for the new stratum and risk having issues with it before pools support it.
And will pools support it if everyone is still using V1? And it just keeps looping and never takes off.

Take a look at IP6, been around since 1995 you know pre internet boom and it's adoption is still minuscule. Heck some cable ISPs are still supplying modems that do not have V6 enabled.....

-Dave

[mikeywith]

Been thinking about this since you posted it a week ago. I keep looping around. Are miner manufacturers going to create new binaries for the new stratum and risk having issues with it before pools support it.
And will pools support it if everyone is still using V1? And it just keeps looping and never takes off.

Take a look at IP6, been around since 1995 you know pre internet boom and it's adoption is still minuscule. Heck some cable ISPs are still supplying modems that do not have V6 enabled.....

-Dave

It's all about the 'consumers', the backbone of mining pools is the miners' owners, as a pool operator you would want to attract as many miners as possible, maybe at some point miners might want to get involved in the blockchain making by actually including certain transactions (maybe those transactions blocked by certain governments), or they want more privacy (encryption) or any other feature that Stratum V2 offers, and given that Slush pool has already implement it, it gives them a headstart and my guess is -- most other pools will follow it.

manufacturers are a different story, look at Bitmain after all these years, a few devs managed to create A LOT better firmware versions and Bitmain did not bother to improve their old shit, and given that compensation is almost nonexistent I honestly don't know what would be the spark for Bitmain and their rivals.

The same goes for IPv6, both of these improvements are not urgently needed and thus people are not rushing to use them, so you are right, Stratum V2 could very well be the next IPv6 for a good while.

[kano]

...
manufacturers are a different story, look at Bitmain after all these years, a few devs managed to create A LOT better firmware versions and Bitmain did not bother to improve their old shit, and given that compensation is almost nonexistent I honestly don't know what would be the spark for Bitmain and their rivals.
...

I see this a lot but wonder what specifically it is that makes these 3rd party firmware that takes a percentage of your mining A LOT better?

The statement you've made is quite specific: "A LOT better" and you go as far as saying that Bitmain should "compensate" them ...

I will add that my releases of S1,S2,S3 allowed you the change the frequency coz the BM driver let you do it anyway - wasn't really that big a deal at all actually.
The biggest issue was the risks of damaging the miners and then people going back to Bitmain to complain about it.
I put a warning on the web page and in the selection list.
I'd guess people damaging their miners then wanting BM to fix them would be a reason Bitmain stopped including that in their versions.

These 3rd party firmware like to go on about how fast they can go ... but if it damages your miner ... bad luck.

[philipma1957]

After market firmware that charges 2-3% is hugely better if you need to run higher in the winter and lower in the summer.

Ie an s19pro has one speed.

where I live it is typically 10c or 50f or less from Nov 1 to April 1 so running the s19pro at 110th makes sense.

but in the summer say June 1 to Oct 1 temps are  75f to 105f or 20c to 40c so running the same s19 pro at

90th on braiins is worth the 2-3% fee as the gear wont over heat.

But I dont need braiins when it is cool and fuck paying them 2-3% from Nov 1 to April 1

So yeah braiins is a lot better and worth paying 2% during hot weather.

Then go to the mine and pull the sdcard to run the s19pro fee free for five winter month’s

obviously the fall and spring months may need sdcard or no sdcard since weather can be off.

to sum up
when i run my s19pros
the braiins firmware is not a lot better for me in the winter.

for sure

the braiins firmware is a lot better for me in the summer

[mikeywith]

I see this a lot but wonder what specifically it is that makes these 3rd party firmware that takes a percentage of your mining A LOT better?

Using custom firmware my miners run cooler at the same hashrate and consume less power, the auto-tuning per chip allows such a thing to happen, while on stock firmware the best you can do is find some stock version that might allow you to change the voltage and/or frequency of each board alone which doesn't bring the miner efficiency nearly as close, just a random example would be S9s running at anywhere between 70-75w per TH which you can't possibly achieve without tuning every chip differently.

You can even have two different profiles to switch between automatically during day/night, this is very helpful for people who pay different power rates at different times of the day, or for fall/spring seasons when it's warm around noon only, on S9 I can set 14TH mode from 6PM to 12PM and 12TH for 12PM to 6PM , both modes will be more efficient than the stock.

The auto-tuning and profile switching is something I personally use which is why I talked about them in detail, the rest of the features such as immersion cooling, overclocking, full SSH access, and whatnot can be found on that custom firmware webpage.

I also can't tell why other people use custom firmware, but I'd assume most of them do it for the same reason, I'd imagine that by now most people who still run S9 or any of the 17 series probably run it on custom firmware since the stock settings are not profitable for most power rates out there.

With stock firmware, if you leave one dead hash board on the miner it will cause issues such as rebooting, taking forever to boot as it has to test to the dead board a dozen times, it's even worse on Whatsminer as you can't just unplug a power cable and you are forced to take the bad hash board out of the miner which messes with the airflow, some custom firmware have the option to programmatically disable the bad hash board so the miner will treat it as "unplugged", I had a long conversion with Whatsminer's support admin regarding this, and despite the fact, that people report this issue to them all the time, they still did not solve it yet!.

The lack of features isn't just Bitmain's issue, even Whatsminer's firmware is pretty basic, it's somehow better than Bitmain, but you can only run the miner on 3 modes, high, normal, and low, the low is just TOO low, so for M20s you go from 68th to 40 something TH, and for the high, you go from 68th to probably 80 something, so you have a total of 3 profiles to choose from, which is why I decide to let my Whatsminers run a bit hot during the summer because I can't afford to lose nearly 40% of the hashrate, I could really use a profile that sets the miner at 10% less hashrate, anywhere around 55th for M20s or 45th for M21s would really do me good in the summer and around noon for except for winter season.

But well!, they don't want to bother with that, so when and if someone makes custom firmware for Whatsminer, I would certainly want to use it.

Stock firmware doesn't even allow you to change the fan settings, let alone do something else with your miner.

Personally, I use AwesomeMiner distribution of Vnish, I can either pay 1.8% and still have to buy AwesomeMiner license or pay 2.8% and get AwesomeMiner for free, I have already bought a license for 200 miners, which I can run at 1.8% and the rest I pay 2.8% fees including license, if the stock firmware allowed me to do the things I can do with this custom firmware, I would most certainly want to keep the fees, but until that happens, I'll continue to use custom firmware.

[NotFuzzyWarm]

Going a bit OT here but ja, why Bitmain, microBT, and since their IPO even Canaan, don't open up their auto tuning routines for users to tweak some of the operation targets is beyond me. For the record, Canaan used to market their miners as constantly monitoring each chip and periodically tweaking them to a defined performance level. A lot of the more general tweaks are available through the cgminer API that Kano wrote. Canaan's entire code for that tuning & chip performance monitoring as well as of course all cgminer switches supported - including the Secret Sauce drivers for all of their chips used up to the A10 series - is out there in the wild. If the mfgrs are so afraid of the inevitable warranty issues then activate one of the non-resetable flags inside the micro-controlers used to signal intentional operation outside of OEM boundaries so the OEM can check it and also grab a log of settings and take it from there with the user as yes warranty or no.

So... Since example open-source code (from Canaan) exists for auto-tune routines as well as per-chip monitoring/control code, ja, someone should build on that to work with other miners and publish it! Since at one time most miners (except perhaps Bitfury) did and these days still *do* follow the cgminer API at least up to a point, it is not too hard to hack their version of cgminer for the basics of Vcore, fan speed, etc. Heck using Awesomeminer, just pull the list of API commands supported and talk to the miner using ssh/GUI.

The hard and most important part is the chip driver(s) used to access whatever resources said chip has regarding the monitoring/controlling of it. *That* Secret Sauce is why the closed source fee based 3rd party FW exists. Throwing in a decent GUI for tweaking that uses all chip control resources available vs using a more limited command line options switch is nice but also the Marketing hook. Canaan *does* support some tweaks for the A10/11/12/13(?) and have published command line support for that but again, since their IPO beyond that limited bit AFAIK they have not published any of their code.

At least Kano made the s17 driver for the Compac-F open source  

[unicornmangle]

If the mfgrs are so afraid of the inevitable warranty issues then activate one of the non-resetable flags inside the micro-controlers used to signal intentional operation outside of OEM boundaries so the OEM can check it and also grab a log of settings and take it from there with the user as yes warranty or no.

in my experience 1 go round with trying to warranty any miner is enough to deter any new or small time miner from seeking help from a mining mfg. I think if the true goal is everyone mining than just give all the information so people can make educated decisions and remove warranty after X amount of hash done by miner maybe lock changes for X amount of hash time.

I have spent a lot of time trying to figure out how the miners work and almost nobody has the time to do that so i think if you wanted to stiffle mining for new/small timers keep doing exactly what has been done by mining mfgs and all these aftermarket firmwares nobody really wants to run.

[mikeywith]

If the mfgrs are so afraid of the inevitable warranty issues then activate one of the non-resetable flags inside the micro-controlers used to signal intentional operation outside of OEM boundaries so the OEM can check it and also grab a log of settings and take it from there with the user as yes warranty or no.

For a very long time, I thought that was the case, which is why I "excused them" for why trying so hard to lock the miners, but it does not seem to be the case, many people would flash custom firmware on their brand new gears, overclock the heck out of them, and if something goes wrong, they would just revert to stock firmware, send the gear for warranty and Bitmain would happily fix it.

So, it's either Bitmain doesn't know about the existence of custom firmware, or they know, and they just can't be bothered.

What Whatsminer does on the other hand is they would give you the "special" overclocking firmware if you contact them directly, it's not publicly available, so they probably need to attach it to your own mining gear and then automatically take that gear out of warranty.

But we are not even talking about overclocking here, we are talking about underclocking, which is even better for said companies, if people run their miners a lot cooler and at lower clocks, this would result in fewer repair tickets, it almost seems like these companies want their gears to die so they can sell you the next generation, it looks as if they try to ensure that gear won't break during the warranty period at stock but then break right after. . I know that may not be the case, but I can't think of a better reason why won't they allow us to underclock our gears!

I don't think we went way off-topic here (maybe a little), but we are addressing the resistance to change on these manufacturers' behalf. When they won't even add an option to allow users to change fan speed -- we should not expect them to rush into implementing something like Stratum V2.

[DaveF]

If the mfgrs are so afraid of the inevitable warranty issues then activate one of the non-resetable flags inside the micro-controlers used to signal intentional operation outside of OEM boundaries so the OEM can check it and also grab a log of settings and take it from there with the user as yes warranty or no.

Lets start with:
"I didn't do that, and the manufacturer is lying to get out of a repair"

But, beyond that we already see that a lot of miner quality has gone to crap and they don't want to spend the time and effort and development money to do anything but get this generation out the door so they can start on the next.

Also, and this is also probably a large part of it too, what we want is really not important, it's what the big mining operators want. We are an afterthought to generate a bit more money.
If every active miner on bitcointalk bought 10 of the latest bitmain miner, they would have still sold less then Riot Blockchain bought......

-Dave

[philipma1957]

If the mfgrs are so afraid of the inevitable warranty issues then activate one of the non-resetable flags inside the micro-controlers used to signal intentional operation outside of OEM boundaries so the OEM can check it and also grab a log of settings and take it from there with the user as yes warranty or no.

Lets start with:
"I didn't do that, and the manufacturer is lying to get out of a repair"

But, beyond that we already see that a lot of miner quality has gone to crap and they don't want to spend the time and effort and development money to do anything but get this generation out the door so they can start on the next.

Also, and this is also probably a large part of it too, what we want is really not important, it's what the big mining operators want. We are an afterthought to generate a bit more money.
If every active miner on bitcointalk bought 10 of the latest bitmain miner, they would have still sold less then Riot Blockchain bought......

-Dave

realistically in a perfect world I have 800kwatts + 400kwatts + 100kwatts

total is 1.3megawatts x 80% is 1.04 megawatts that is 300 s19 xp units. Is the most I can ran and I do not have the capital to get there.

 But if I was there I certainly want to drop them 10-15% for the summer. I want to do it sitting in my house 77 miller from mines a+b and 115 miles from mine c

Right now braiins would do the trick and bitmain offers 1 speed for that gear which does not do the trick.

300 ip's at best is 300 x 5200 = $1,560,000 for the xp's and I want to shift them lower in hashrate once temps go over 80f 90f 100f  that will happen on many days from April to Oct 1

I also rather run fans at 90% not 100% so if I have to run a 141 unit at 130 to be safe at 90% it is worth it. As driving from my house to locations that are 115 miles away to replace fans would suck.

[mikeywith]

300 ip's at best is 300 x 5200 = $1,560,000 for the xp's and I want to shift them lower in hashrate once temps go over 80f 90f 100f  that will happen on many days from April to Oct 1

Ya, that's the issue with these miners, they put way too many chips in them, and clock them pretty close to their max so they would only run pretty cool when temps are acceptable, but then depending on where you put your miner, there would be at least a few weeks of pretty hot weather, it makes no sense to spend tens of thousands of dollars on cooling the farm based on a temp of 40c when that 40c only happens for a few weeks.

To solve that issue you can do

1- Shut down your miners when the outside temp is high.
2- Spend thousands of dollars on cooling which you won't be using for the majority of the year.
3- Underclock your miners when it's too hot, so they are not totally off and they are not running hot, and you have not spent a huge amount of money on cooling that you won't use when the summer is over.

It's pretty obvious that the last option is the best option, the custom firmware folks understand the need for such flexibility, they also know that people will rather pay them 2-3% of their hashrate than turn the miner off, they also know that in most cases that 2-3% is much less investing in more cooling, Bitmain did make a someone good firmware for the S9s back then, you could change the hashrate by -+ 1-2 terahash, your use the "Low-Power-Mode" which cuts about 5th out of the default 14th, okay not as rich in functions as custom firmware but was good enough for many people.

Starting from the 15 series onwards, they stopped doing that, and then they made the WORST miner series ever (The 17 Series) which is when custom firmware started to spread pretty fast because many people needed to underclock those crappy miners or mix and match hash boards, I still remember how unpopular the various custom firmware was before the end of 2019, their telegram channels had a few hundred users at most, some probably had no more than a 100, next thing you know, thousands of people became interested in custom firmware, I was expecting Bitmain would do something about it, but they didn't seem to bother.

[cygan]

i am not really familiar with this topic and have not followed this discussion/thread but today i stumbled across the following information about stratum v2
at the following link you can find the repository of Gabriele Vernetti. it contains his personal work, written during the completion of the master's degree in computer engineering (cybersecurity) at the politecnico di torino.

Stratum V2: the next generation protocol for Bitcoin pooled mining

https://github.com/GitGab19/Stratum-V2-Master-Degree-Thesis