BitCrack - A tool for brute-forcing private keys

[bigvito19]

Most likely there will have to be another program to be made to solve the remaining keys left between #64 - #160. I don't see bitcrack solving no more keys, its no longer getting the job done.

[Trancefan]

secp256k1::uint256 CLKeySearchDevice::getNextKey()
{
    uint64_t totalPoints = (uint64_t)_points * _threads * _blocks;

    return _start + secp256k1::uint256(totalPoints) * _iterations * _stride;
}

Can I add some creative counting to the function? For example, ×2 private key, ×3 private key, or using fibonacci sequence.

[NotATether]

secp256k1::uint256 CLKeySearchDevice::getNextKey()
{
    uint64_t totalPoints = (uint64_t)_points * _threads * _blocks;

    return _start + secp256k1::uint256(totalPoints) * _iterations * _stride;
}

Can I add some creative counting to the function? For example, ×2 private key, ×3 private key, or using fibonacci sequence.

The value of _start + secp256k1::uint256(totalPoints) * _iterations * _stride is going to be the next private key so if what you are trying to accomplish is to "skip over" private keys, multiply this result by whatever value you want to use (except you shouldn't multiply _start with this number), and pass an additional argument to getNextKey() that is a pointer to struct that contains the next number in the sequence to multiply by. Perhaps also include in there a number that the limit to the number of terms in the sequence to use, after which it wraps around.

You could even make this number a modulus for excessively large sequence terms to be divided into.

Code:

typedef struct nextkey_state {
    uint64_t index;
    uint64_t limit; // or uint64_t modulus
} nextkey_state_t;

secp256k1::uint256 CLKeySearchDevice::getNextKey(nextkey_state_t *state)
{
  // ...
}

And then of course update all function calls to use this parameter.

[mrxtraf]

https://youtu.be/0unf-S509gE
One more tools. 

[dextronomous]

https://youtu.be/0unf-S509gE
One more tools. 

you again posting link to your own ecdsa supra ? gaylord.kuhic26@contbay.com

ok you smart. got a program many want. 50k now 20k on satoshidisk.
just figured that one out?

i asked if like many could get a copy. but hey who's nice.
 

[NotATether]

https://youtu.be/0unf-S509gE
One more tools. 

you again posting link to your own ecdsa supra ? gaylord.kuhic26@contbay.com

ok you smart. got a program many want. 50k now 20k on satoshidisk.
just figured that one out?

i asked if like many could get a copy. but hey who's nice.
 

Yeah, I just watched the video and from reading that satoshidisk link it looks like he's trying to sell it for 0.26BTC. I also took a screenshot of the video:



If you look closely at the top you can see that he's running a PHP script from Linux or MAC, that connects to Google BigQuery (a cloud computing service) to run it's brute forcing. This raises another red flag, since by paying this guy $10K you get a script that can become worthless if he shuts off the cloud computing services and optionally runs another instance of the service at some other IP address. Which is a shame because it actually looks like a promising tool, which can search more than 160 bits space (I'm guessing the "61 160" at the top right is the bit range).

[dextronomous]

me was thinking that same i guess,
would love a program that moves around the cursor positions and tell me where the right
one should go to.

[PawGo]

How do you know that what you see is a real result of calculations, not just a movie?

[bigvito19]

It seems like a good program that we all could put some use to. I see he was charging at least $50k for this a year ago.

[MeBender]

Yo why is anyone here talking about buying or selling software when JeanLucPons has already released open-source GPU kangaroo software?

https://github.com/JeanLucPons/Kangaroo

[WanderingPhilospher]

How do you know that what you see is a real result of calculations, not just a movie?

I'm not sure what it is even doing/showing....calculating the pubkey via binary to hex?
I will make one as well, will anyone be interested in buying it? I will sell it for half of what he is selling for 

[renedx]

My base logic is always, if someone would know/got a better way, why would you sell it. There is enough puzzle bitcoins for grabs if the tool would be real.

Edit:
Ah, his tool is a database of wallets + pub keys.
See https://bitcointalk.org/index.php?topic=5307531.msg56042517#msg56042517

[NotATether]

So NVIDIA just nerfed their new GPUs for mining: https://www.xda-developers.com/nvidia-cmp-cryptocurrency-mining-processor/amp/ says that ethereum hashrate on the 3060 has been nerfed by 50% by some firmware inside the GPU that eerily reminds me of hardware-based DRM.

What effect do you think this will have on the speed of Bitcrack and friends? (Yes I know Bitcrack doesn't even work on RTX 30, I am working with someone right now to lease a Tesla GPU from AWS which should be finished one of these days so I can finally start thrashing that monster bug  )

It looks like they are detecting if certain hashing functions are being used in CUDA but I'm not sure which. How do they do that anyway, since there is no builtin SHA or elliptic curve or RIPEMD160 libraries in CUDA and these all are user-written?


In other news they're also releasing a GPU-as-a-miner kind of card. I wonder if Bitcrack will work on those without modifications or if these can only mine Ethereum.

[EXXXXO]

Hello guys,
Is there a way to stop the work and continue later at this point?


https://i.imgur.com/bAiYIFc.jpg

[MeBender]

Hello guys,
Is there a way to stop the work and continue later at this point?

Holy crap thats some setup you have going! that's like what 4-5 v100s?

But anyway, if you want to save the point to start again later you can take the number in the brackets, in your example  the number would be 812,819,340,591,104 and covert that from decimal into hexadecimal here: https://brainwalletx.github.io/#converter which in the example would output 02e34140100000, you would just use that as your keyspace start point next time you want to run it.

It looks to me though that you are trying to crack a single address by running through the entire 2^160 keyspace? Would you not be better off trying to crack challenge address #120 with pollard kangaroo software? https://github.com/JeanLucPons/Kangaroo

[EXXXXO]

Hello guys,
Is there a way to stop the work and continue later at this point?
https://i.imgur.com/bAiYIFc.jpg

Holy crap thats some setup you have going! that's like what 4-5 v100s?

But anyway, if you want to save the point to start again later you can take the number in the brackets, in your example  the number would be 812,819,340,591,104 and covert that from decimal into hexadecimal here: https://brainwalletx.github.io/#converter which in the example would output 02e34140100000, you would just use that as your keyspace start point next time you want to run it.

It looks to me though that you are trying to crack a single address by running through the entire 2^160 keyspace? Would you not be better off trying to crack challenge address #120 with pollard kangaroo software? https://github.com/JeanLucPons/Kangaroo

Thanks for your help MeBender.
I am running BitCrack on an ASUS ROG Zephyrus G14 notebook,
NVIDIA GeForce GTX 1660 Ti 6GB GDDR6 Dedicated Graphics

pollard kangaroo does not work on my system. Unfortunately I have not found a way to install the CUDA SDK 10.2. I'll try again this weekend.

(sorry for my broken English)

[NotATether]

I am running BitCrack on an ASUS ROG Zephyrus G14 notebook,
NVIDIA GeForce GTX 1660 Ti 6GB GDDR6 Dedicated Graphics

pollard kangaroo does not work on my system. Unfortunately I have not found a way to install the CUDA SDK 10.2. I'll try again this weekend.

CUDA toolkit 10 won't recognize a 1660Ti because that's a Turing card and support for those were added in CUDA 11.

What kind of error did Kangaroo make? Did it look like a driver or CUDA related error or did you find an actual bug?

[WanderingPhilospher]

I am running BitCrack on an ASUS ROG Zephyrus G14 notebook,
NVIDIA GeForce GTX 1660 Ti 6GB GDDR6 Dedicated Graphics

pollard kangaroo does not work on my system. Unfortunately I have not found a way to install the CUDA SDK 10.2. I'll try again this weekend.

CUDA toolkit 10 won't recognize a 1660Ti because that's a Turing card and support for those were added in CUDA 11.

What kind of error did Kangaroo make? Did it look like a driver or CUDA related error or did you find an actual bug?

I am running a 1660Ti on CUDA 10.1 (haven't tried on CUDA 10.0); Those cards came out after/during the RTX 20xx series.

[WanderingPhilospher]

Hello guys,
Is there a way to stop the work and continue later at this point?

Also, next time (before you run the program again) you can add the " --continue somefile.txt  "flag and the program will save your progress every 60 seconds.

[renedx]

Hello guys,
Is there a way to stop the work and continue later at this point?

That's some insane numbers, you sure thats a 1660 Ti? GFX902 is more like a Radeon RX Vega.
Prob. the reason Kangeroo doesn't work for you, that is CUDA only.