Looking for a working configuration for a BTCrecover seedlist token file

[NotATether]

So it did it again in likely the exact same spot of 2.8 billion.

Now I'm trying to work out where to stick the --autosave command in the argument. Could use some extra insight and common sense as my brain is migrating towards my sledgehammer at the moment.

Was your seedrecover invocation using GPU acceleration? Because sometimes a badly written nvidia CUDA program locks up or crashes not just Windows but also Linux systems.

This kind of thing should not happen if it is only CPU-bound, because the worst that would happen would just be a program crash.

[kauai_lost_boi]

Crypto Guide suggested I may be running out of memory and that's why the OS crashes at the same spot. I got 40 gb of memory and 6 disks in a 0 raid array. Can't think of any other reason why it would crash other than Win 10 or anything Microsoft would be the weakest link

[kauai_lost_boi]

So it did it again in likely the exact same spot of 2.8 billion.

Now I'm trying to work out where to stick the --autosave command in the argument. Could use some extra insight and common sense as my brain is migrating towards my sledgehammer at the moment.

Was your seedrecover invocation using GPU acceleration? Because sometimes a badly written nvidia CUDA program locks up or crashes not just Windows but also Linux systems.

This kind of thing should not happen if it is only CPU-bound, because the worst that would happen would just be a program crash.

Luckily or unluckily Win 10 wasnt compatible with the Nvidia GPU and I was relying on dual processors and 44 cores to pickup the slack. Given what I'm working with it seems to limp along ok.

I was considering setting up btcrecover on a linux OS since openCL can't take advantage of the outdated GPU, or I may run it on my gaming computer for a week or 3 if it will get better results.

[NotATether]

Crypto Guide suggested I may be running out of memory and that's why the OS crashes at the same spot. I got 40 gb of memory and 6 disks in a 0 raid array. Can't think of any other reason why it would crash other than Win 10 or anything Microsoft would be the weakest link

Have you ever measured the memory usage from Task Manager? It's true you'll get a blue screen if you finish both your RAM and swap (and the swap itself is quite small ~8GB), but normally the system's supposed to slow down to a crawl first as your swap is filled and you're out of RAM.

Luckily or unluckily Win 10 wasnt compatible with the Nvidia GPU and I was relying on dual processors and 44 cores to pickup the slack. Given what I'm working with it seems to limp along ok.

I was considering setting up btcrecover on a linux OS since openCL can't take advantage of the outdated GPU, or I may run it on my gaming computer for a week or 3 if it will get better results.

I don't think Windows 10 will work with GeForce 8xxx and 9xxx if that's what you have, all newer cards should be recognized.

That may also explain why they don't have decent OpenCL support for them  

[kauai_lost_boi]

Crypto Guide suggested I may be running out of memory and that's why the OS crashes at the same spot. I got 40 gb of memory and 6 disks in a 0 raid array. Can't think of any other reason why it would crash other than Win 10 or anything Microsoft would be the weakest link

Have you ever measured the memory usage from Task Manager? It's true you'll get a blue screen if you finish both your RAM and swap (and the swap itself is quite small ~8GB), but normally the system's supposed to slow down to a crawl first as your swap is filled and you're out of RAM.

Luckily or unluckily Win 10 wasnt compatible with the Nvidia GPU and I was relying on dual processors and 44 cores to pickup the slack. Given what I'm working with it seems to limp along ok.

I was considering setting up btcrecover on a linux OS since openCL can't take advantage of the outdated GPU, or I may run it on my gaming computer for a week or 3 if it will get better results.

I don't think Windows 10 will work with GeForce 8xxx and 9xxx if that's what you have, all newer cards should be recognized.

That may also explain why they don't have decent OpenCL support for them  

Yeah, I got 40 GB of ram. That is a big 4 (four) 0 (zero) F O U R T Y GB of ram and it's still crashing.

So I'm gonna take a small break, let my head recover from bashing it into a wall so many times, then I will install ubuntu 20.04 then install all the needed packs for BTCrecover to work.

Thanx for all the help you guys have given me and hopefully when I return it will be with news of progress and potentially leading to a happy ending for this story.

[kauai_lost_boi]

Happy Summer and hope everyone is enjoying some modicum of festivity in frivolity. It's been a minute or 2 since my last post and I am back with a working version of BTCrecover in a Linux OS environment.

After initial tests it seems to run slower than molasses over a witches nips in January but if I gotta sacrifice speed for stability for a time, I'll try once more before I run out and buy one of those new zen 3 rdna2 radeon 6000 gaming rigs to run btcrecover

As Oeleo worked out, I significantly need to decrease the amount of entropy left in my seedlist. Still trying to get a working configuration. My latest hangup is it completes too quickly and I have no confidence my seedlist token file was configured right

I still got some time this week to figure out how to modify my arguments for command line. Let me know if anyone has some working examples or insights.

[o_e_l_e_o]

My latest hangup is it completes too quickly and I have no confidence my seedlist token file was configured right

Have you tried using the --listpass command to have btcrecover print all the seed phrase combinations it is trying rather than actually trying them?

BTCRecover can also print the seeds that will be tested via the --listpass command, something that can be useful for debugging your tokenlist

Do you want to share you current tokens file (with the words obfuscated, of course) for us to take a look?

[kauai_lost_boi]

Thank you oeleo for everything you've helped with. It's good to see you still rockin and rollin with me on the forum

I didn't even know what the listpass command did let alone how to implement it. I was worried creating too many things in the argument would contribute to bogging things down let alone going through to another crash. I'll look into this

My latest hangup is it completes too quickly and I have no confidence my seedlist token file was configured right

Have you tried using the --listpass command to have btcrecover print all the seed phrase combinations it is trying rather than actually trying them?

BTCRecover can also print the seeds that will be tested via the --listpass command, something that can be useful for debugging your tokenlist

Do you want to share you current tokens file (with the words obfuscated, of course) for us to take a look?


This is a mix of what I need and what I was trying to get to work

^1^1stwordlocked
^2^2ndwordlocked
one word is known out of 4 spots
7th and 8th words are known but unsure of order
^9^9thwordlocked
^10^tenthwordlocked
^11^eleventhwordlocked
^12^lastwordlocked
I believe I'm supposed to include 3 lines of all mnemonics here to fill the unknowns


I can try to narrow things down more but if I wanted to take the slow boat with a more assured approach for my needed results, I need the wiggle room at the sacrifice of time

[o_e_l_e_o]

Yeah, I think the best way to do thisnis going to be to lock words 7 and 8, and then lock the last known word to slot 3. Then, as you say, you need to include three lines of every word on the word list. This should give 8.6 billion possibilities. Run that to completion, and if no success then move the locked word from slot 3 to slot 4 and repeat. Do the same again moving it to slot 5, then to slot 6. Then reverse words 7 and 8 and do another 4 runs with the last word in slot 3, 4, 5, then 6.

I think this is the best way of approaching this without making your tokens file overly complicated, and also means if you crash or freeze you don't havenl to repeat everything from the very start.

[kauai_lost_boi]

Right now I'm running the routine with the "best guess" option. I was intending to see how stable this Linux OS is with BTCrecover. Originally it read at 48.5 kp/s which was half of what the Win 10 OS was putting out, but in this instance numbers did lie

I've been running it for 14 hours and I'm at 2.5 billion results. (around 200 million an hour) Thats a day and a half shorter than the Win 10 OS and it seems much more stable. I've witnessed ZERO hangups and it seems to keep pluggin and chuggin along. Once I cross 3 billion, I will know I got a better approach model. Anyone know how long this routine would run?


In regards to the seedlist token file, I know I can lock the loose known word to the third or fourth position. I am 90% certain I can lock positions for the others. And as you recommended, I can play mnemonic jenga and token tetris later if they don't play nicely.

Hope I get results... My power bill is gonna suck this month  

[kauai_lost_boi]

So I hit a major milestone. I just hit 3 billion results at the 17 hour mark using the BTCrecover configured Ubuntu 20.04 OS with absolutely no issues at all. I'm not sure what the issue was with the previous attempt using Win 10 that made so many issues and ultimately making it crash.

The Linux method seems to have less jinxable bits to it. Gonna let it run till it gives me a reason to do my happy dance or it goes splat in which I've prepared the new token txt file.

[o_e_l_e_o]

Hope I get results... My power bill is gonna suck this month  

You certainly deserve them given how much work you've put in.

I'm not sure what the issue was with the previous attempt using Win 10 that made so many issues and ultimately making it crash.

Win 10

I think that's your problem right there.

So at 17 hours for 3 billion combos, that's about 180 million an hour. To try 8.6 billion possibilities as above should take around 48 hours. With words 7/8 being swapped, and the last known word in position 3 or 4, that's 4 sets of 8.6 billion, coming in at 8 days (provided everything else in your tokens file is correct). Good luck!

[kauai_lost_boi]

For this shakedown pass I've initially gone with the "best guess" option with the 3 big typos in the argument. I saw the remarkable difference in this stress test and decided to let the thing fly. Just wondering if I'm expecting results by 9 billion without the word locks or if i'm gonna be waiting for weeks and should stop it now. After 31 hours I'm standing at 6.15 billion

I've got the seedlist token file for the next attempt ready and waiting.

[kauai_lost_boi]

So it's been 48 hours and I'm at 8.4 billion. I'll leave it overnight and check for replies in the morning. Maybe when I rise for the day I'll have some lucky surprise to greet me. If that happens i'll be like a kid again giddy for a new season of Saturday morning cartoons and a big bowl of sugary cereal.

[o_e_l_e_o]

For this shakedown pass I've initially gone with the "best guess" option with the 3 big typos in the argument. I saw the remarkable difference in this stress test and decided to let the thing fly. Just wondering if I'm expecting results by 9 billion without the word locks or if i'm gonna be waiting for weeks and should stop it now. After 31 hours I'm standing at 6.15 billion

Did you fill in 3 random words for the 3 words you had missing and then just run the basic recovery without a tokens file? If that's the case, then your other 9 words aren't locked, and so it will try replacing any 3 words in your seed phrase with others. Since there are 220 different ways to pick any 3 words out of 12, and there are 2048³ possible combinations for those 3 words, then that would result in somewhere around 1.89 trillion combinations. At your current speed of 178 million per hour, that's going to run for 442 days. That doesn't even include other "typos" it might try, such as words being swapped or in the wrong order.

[kauai_lost_boi]

At your current speed of 178 million per hour, that's going to run for 442 days.

Welp, had to kill that process because I'm likely gonna have to pay the power bill in the next year and a half. But I did manage to get 9.6 billion results before I closed the terminal.

Now I'm still dealing with the seedlist token file and it doesn't wanna make things easy. I'm still convinced I'm not using this thing right or creating a correctly formatted token file.

My current argument is

Code:

python3 seedrecover.py --no-dupchecks --mnemonic-length 12 --language EN --dsw --no-eta --typos 3 --big typos 3 --seedlist ./seedtoken.txt

I then get the GUI pop ups

I cancel the wallet file

I select Bitcoin Standard BIP39/BIP44

I cancel Master xpub key

I cancel Addresses and run off the address DB

I chose the default 10 addy limit, then it runs




I've gone through all the different variations of what we've all discussed for the token file and it's giving me the 'seed not found, sorry' almost instantly, even when I remove the ^#^ in front of the word.

I know I got a winning approach, even if it might mean sitting on the thing for a weekend or even a week, but I don't believe some things are corresponding correctly.

[o_e_l_e_o]

My current argument is

Code:

python3 seedrecover.py --no dupchecks --mnemonic length-12 --language EN --dsw --no-eta --typos 3 --big typos 3 --seedlist ./seedtoken.txt

Firstly, you have a typo. It should be --mnemonic-length 12. Your dash is in the wrong location.
Secondly, you don't need the --typos 3 or --big typos 3 arguments when using a tokens file. Your tokens file should be configured in such a way that it contains all the possible combinations to try. You can use these I believe, but they will exponentially increase your search space.
Thirdly, and the reason you are getting the "seed not found" error, is because you are using --seedlist instead of --tokenlist. The seedlist command tries every line in the text file as an individual seed, whereas the tokenlist command treats each line as an individual word in a potential seed phrase.

[kauai_lost_boi]

Firstly, you have a typo. It should be --mnemonic-length 12. Your dash is in the wrong location.
Secondly, you don't need the --typos 3 or --big typos 3 arguments when using a tokens file. Your tokens file should be configured in such a way that it contains all the possible combinations to try. You can use these I believe, but they will exponentially increase your search space.
Thirdly, and the reason you are getting the "seed not found" error, is because you are using --seedlist instead of --tokenlist. The seedlist command tries every line in the text file as an individual seed, whereas the tokenlist command treats each line as an individual word in a potential seed phrase.

Glad to have you with me this morning. The only silver lining to not being able to sleep. So I fixed the code. I free typed it and fatigue is my excuse for mistakes.

I believe Cryptoguide also said the typos would be redundant. Thank you for reminding me. Been 2 1/2 months and still trying to jumble everything together.

So the "token list" now looks like this


1^1^lockedword
2^2^lockedword
3^4^lockedword (switching between position three and four)
4^7^lockedword
5^8^lockedword
6^9^lockedword
7^10^lockedword
8^11^lockedword
9^12^lockedword
10 abandon ability etc
11 abandon ability etc
12 abandon ability etc



The argument now looks like

Code:

python3 seedrecover.py --no-dupchecks --mnemonic-length 12 --language EN --dsw --no-eta --tokenlist ./seedtoken.txt

I just hit enter and BEHOLD, it's running and counting!!! It's 5.30 in the morning, is it too early to celebrate this tiny victory with a bottle of Guinness...? 

[kauai_lost_boi]

So it's been 5 hours and for whatever reason it's seems to be parsing at 20.1 kp/s and only at 360 million results. At least it's working.

Even with 24 worker threads, 44 cores and 6 striped disks, I may need to get a 'bigger boat'.

Went from 170 million to 7.2 million an hour. Sacrificing speed for semblance and stability indeed 

[kauai_lost_boi]

So I'm happy I got the token file working but I'm mystified as to why it's going so damn slow. 13 hours in and I'm just about to crack my first billion.

Patience is definitely virtual and good things come to those who have high bitrate. Might be time for that Guinness, in a nice hot bath and a good Terry Pratchett book while I meditate on good things