Braiins OS & Braiins OS+ custom ASIC firmware: optimize performance & efficiency

[Artemis3]

So how these third party firmware give u a pretty close reading ?

I tested the reading vs physically taking the reading and they get pretty close .  +-20watts in total consumption.

It is true that the hardware has no direct means of measuring this. My guess is that they took some measures manually at different speeds and voltages and make an educated guess.

[Keezasa]

Is the Antminer S9se supported?

My board does not have the 4 jumpers.
https://docs.braiins-os.org/s9.html
It only has one jumper by the fan connectors.
It states jumper 1.
I NEED fan control.

So I have discovered the SE is the same as the S9k.
No Four jumpers. only the one jumper. which seems to do jack.

I wish I has seen this post from Artemis3 before I "upgraded" to sh1tville:

Moral of the story: S9 owners DO NOT upgrade firmware to 2019 version, you will lose SSH access!!!

I wrote that for a reason. Still, you could use an SD card to flash it back using that convoluted T9 firmware method, or simply boot Braiins OS and install that instead.

I don't think S9s should be running any other firmware at this point, unless you have the ancient controller.

With Braiins OS you can improve the efficiency from 100w per tera hash down to around 80w. This alone increased the lifetime of the units against Bitmain's wishes, which is why they responded by locking down ssh access since 2019.

[Artemis3]

The SE (and K) are not supported (don't work) with BraiinsOS yet. I think i left various links in that thread pointing for possible methods to regain SSH, but those models came with SSH access disabled from factory because they are 2019 models. I believe you can still control the fans using some html trickery with the factory firmware.

The bOS team plans to add support to more hardware such as the S17 after the next release.

One of the reasons they rewrote cgminer from scratch in rust (called bosminer) was precisely to ease adding more hardware.

[Keezasa]

The SE (and K) are not supported (don't work) with BraiinsOS yet. I think i left various links in that thread pointing for possible methods to regain SSH, but those models came with SSH access disabled from factory because they are 2019 models. I believe you can still control the fans using some html trickery with the factory firmware.

The bOS team plans to add support to more hardware such as the S17 after the next release.

One of the reasons they rewrote cgminer from scratch in rust (called bosminer) was precisely to ease adding more hardware.

Thank you for pointing me to the HTML trickery. Works like a gem. Pain to do every time but better than nothing.
I hope an alternative firmware comes out soon.

[Biffa]

So any proof that a miner running Braiins has found a BTC block yet?

[Seph213]

Any template to follow when it comes to tuning the hasboards on brain OS , i know boards are snowflakes but there's gotta be some kind of basic rules they follow ? Other third party firmwares would auto tune it for me .

Now with BOS i have to tune them myself and wondering if there's any pattern/logic that works with this .

i been able to tune it properly, its just some boards are not so simple as others and i feel that i'm maybe missing out on efficiency.

For example i have this one board that was running at 2200 ghs , 8.5 V at 606 freq . I dropped the freq , left the voltage where it is and was able to get it to mine at 3500 + .

Other boards regardless of the freq it would not hash to its full capacity unless i increased the voltage .

So what are the rules of this game ?

[Artemis3]

I have written about that procedure in this thread. Everything is manual, and you do exactly that, per board.

First i don't like going over 9v some people start at 8.7v, then pick a default speed (such as 600). Efficiency vs overclocking path differ, for efficiency start lowering the voltage until it stops hashing or there are too many errors expected hashrate differs too much.

For overclocking once you do this you try one step speed increase and then very slowly voltage increase until everything is normal.

If you see it wanting more than 9v, you know you are getting in the red zone, consider stopping. Also overclocking might need you to change temperature protection.

Everything is manual and you can easily fry your boards by using wrong values. No hand holding (also, no idiotic calibration going wrong in winter).

As you might guess, finding the optimal values take several cgminer resets. So yes, its for the true hobbyist.

Remember each board is different, and when air cooled the middle board gets less ventilation. Use this knowledge to your advantage.


Also next version is coming, its currently getting beta tested.

[Seph213]

Next version will have auto tune ? they should add the estimated power consumption per board .Any eta as to when it'll come out ?

[MoparMiningLLC]

maybe it was posted somewhere here? if so, I apologize. but can anyone show proof of anyone using this FW that has found a block?

[NotFuzzyWarm]

maybe it was posted somewhere here? if so, I apologize. but can anyone show proof of anyone using this FW that has found a block?

So far I have not seen hide nor hair of any proof that Braiins - or any other non-OEM firmware -finds BTC blocks. Considering that it is Slush who is behind the FW and no doubt there are a lot of Braiins users on Slush one would think that it would be very very easy for them to post proof it finds BTC blocks...

But so far <sound of crickets>

[Artemis3]

As time passes, less and less gear will ever find a block in their lifetime. Best you can do is mine some sha256 altcoin and see for yourself or check the source code, it's not like its hidden like the others. Spot anything that would prevent it from finding a block? No? Then where is your proof it doesn't?

Or, you can clone Bitcoin, make your private blockchain for testing and point it there. Are you up to the challenge?

Now go take your silly unfounded misconceptions to the only pool that doesn't like this Free and Open Source firmware, the rest don't care.

TL;DR: Prove that it doesn't or just shut up. Go take your unsubstantiated suspicions to the closed source OEM and modded firmwares, why are you two not challenging every single one of them in their own threads?

I know for a fact a miner with a large bOS install, as typical large miners do, they go with ViaBTC so there. It is false that people use this firmware with just Slush. It is false that Slush only has bOS users, but only the pool operators can tell so go ask them in their Slush thread. The next version, however might be more interesting for large miners to use with Slush (or any pool implementing v2) due to the bandwidth savings and added security among things.

This firmware does NOT use a default pool, unlike what Bitmain does with Antpool. Feel free to use ANY(*) pool with this zero fee dev, Free and Open Source firmware. Audit it, compile it, install your own audited compiled version. Can you do that with OEM? No, because none of them releases the code like bOS does.

Your animosity against bOS is suspicious to say the least, its getting in the realm of FUD.

(*)Except Kanopool, stay away from it.


Seph213: I don't know if it will have auto tune, or power guesstimate. I know it is the first release of bosminer replacing cgminer, which means Stratum V2 support and probably bugfixes (such as no more reordering hashboards). After that release next version is aiming to support newer gear like the S17, that's all i know so far.

[MoparMiningLLC]

i think what he is asking as are a few others - those that use this should have done what you suggested and tested it and shown the results - we are only asking for verification. There have been and are a lot of FW out there that does not and has been proven to not find blocks - flashback to slush and genesis mining. Slush let them run a large qty of hash (think around 300 PH) using a custom FW and paid them - when it was identified that their FW does indeed not find blocks, slush still paid them.

we all want more out our miners but we also want miners that are capable of finding blocks at the end of the day.

as for challenging the others, we have asked - no one answers or the posts get deleted - or we get told to prove it doesnt work. The proving should be the responsibility of the one pushing the fw. I have no animosity nor have my posts shown any so my guess is you mean that towards Fuzzy.

edit: and if it has been shown/proven and I missed it, I apologize and ask that I be pointed to the right thread/post.

[Biffa]

[...]

Riiiight... so you are confirming, there is no proof that this firmware finds blocks.

If you are a supporter of bitcoin then you should be very concerned about this. I don't understand why you sound so butthurt about it. Your not getting paid to shill the software so why the angry tirade over an honest and fair question that EVERYONE should ask.

If everyone on a pool ran firmware that didn't find blocks then that pool would NEVER find a block. Surely that's a concern. Or what about if only 25% of people ran firmware that didn't find blocks, that would only lower the pools block finding ability and earning capability by 25%.

We are fundamentally curious about it, I'm hoping it does find blocks, I'm hoping someone can come and show me that it finds blocks because then no pool will prevent its use.

[NotFuzzyWarm]

As Stryfe said - we have asked for proof in the other FW threads as well. No responses at all or deleted. I will say that it is heartening to at least have a response from you even if it is rather negative... The folks behind Braiins certainly do deserve Kudos for it being Open Source complete with the availability of all source code as required per the cgminer license. You are the only aftermarket firmware group do do that  

Regarding

Can you do that with OEM? No, because none of them releases the code like bOS does.

Mostly right but not entirely accurate. Up to Canaan's A10 series their code is freely published in Github repository. From A10 on up - not sure since they have completely revamped their firmware & software but odds are it *is* available. Then again, since Canaan (up to the A10x) has the cgminer More Options field in their GUI it (was) simplicity in itself to tweak their miners to your hearts content which negated any need for aftermarket firmware...

Never said that Braiins is only used on Slush (and yes it is banned on Kano), was just saying that with Braiins being part of the Slush family, it should be easy for you folks to check how it performs.

My point was that Slush (and other pools) can easily log what FW is in the miners. As they are the ones behind Braiins they should have a vested interested in how it performs and are in the best position to verify it has no problems. I know that Kano periodically checks what FW miners are using because Kanopool (and all other small pools) does not have the hash rate to risk *any* chance of missing a block due to a bug. If the other small pools don't care - well, that's entirely up to them and is just another reason I don't mine elsewhere.

 For pool operators, when a miner finds a BTC block, query the logs to see if it runs Braiins, if it does - let folks know about it. On the actual miner side - especially large farms - the cgminer API has a block found function: Setup your monitoring software to flag it. If a non-oem miner finds a block - and any large farms should find several in a very short period of time - lets folks know! I would think that large farms should not only be checking their hash rate vs income but also be very interested in the "how many blocks did WE find and which miner did it" metrics.

Oh, and yes, the chance of any one single piece of hardware finding a block has always been slim and barring a drop in diff will just keep getting harder. But since 2014 I've found 15 blocks and over the course of 125 miners ranging from S1-S9, an uber rare AMT A1 miner circa 2014, several Avalons from 721-841, and Whatsminer M10's at least 1 from each family has found a block. I have an R4 that has found 2. The only miners I have that have not yet found a block are 2x A941, and 5x A10's. I currently run 25 miners and found Kano's last block with an A841 cluster of 5. Point is that given a large sample of miners running and provided they bother to check, a person/company *will* find blocks with their hardware giving them (and us if they share) data as to what works and what does not work.

Again, I have no animosity towards the Braiins devs and do respect you for fulfilling the cgminer license. But as an example, in the Genesis mining kerfuffle obviously shares from them were accepted by Slushpool as normal and yet there was a precipitous and lengthy drop in the farms find rate due to a bug in the proxy software that Genesis was testing. As in despite running over 300PH Genesis suddenly went from several blocks per day to zero for over a month and until other folks started bitching about the pools Luck and their income dropping apparently no one noticed something was wrong somewhere...

I and others would just like to see proof that nothing was inadvertently broken.

[philipma1957]

 I have 1 s9 with two boards running braiins  this is my only sha-256 asic miner that does not run stock firmware.

One of the reasons I don't have more units mining on this firmware is the no proof issue of blocks.

I have about 2 ph of sha-256 mining on viabtc pps+ it would be nice to see some blocks hit proof for any aftermarket firmware.

[chrysophylax]

I have 1 s9 with two boards running braiins  this is my only sha-256 asic miner that does not run stock firmware.

One of the reasons I don't have more units mining on this firmware is the no proof issue of blocks.

I have about 2 ph of sha-256 mining on viabtc pps+ it would be nice to see some blocks hit proof for any aftermarket firmware.

Then ...

YOU are in a better position with your mining hoard to test this simple but (seemingly) enigmatic question. A single miner will undoubtedly have almost ZERO chance to find any block with BTC, but 2PH could easily solve the problem of answering this question mate.

Whether you WANT to do this is another story

Even at 2PH, the only way that I believe this could be solved is to mine a test block on testnet, considering that the testnet blockchain is exactly (almost) the same as mainnet. If you hit/solve a block there, you are likely to solve one on mainnet, logically speaking.

#crysx

[percy_tc]

Does anyone know if s9 SE will be supported?

[Artemis3]

Does anyone know if s9 SE will be supported?

This model is not yet supported in the expected March release.

[philipma1957]

[...]

when I bring  the s9's off line and start to build them into one and two board space heaters for next winter I will test on solo.ckpool.org

It won't be enough hash  to guarantee hitting a block  , but it could indeed hit one.  While I control 2ph  it would take 13 months to hit a block with it if I have average luck.  So I simply can not run  for 13 months.  Test net is just that test net.  Since diff is lower on test net hitting blocks with a low difficulty proof that you can hit a low diff no more than that.

[milani1387]

hi
i have t1 16 th/s miner.
can i use  braiins frimware

my miner control boards have not memory card socket.