BitCrack - A tool for brute-forcing private keys

[Dexed]

Hey guys. How to run it on RTX 30 or 50? I saw it is a problem, is there any solution?

[SovereignScott]

Hello, first and foremost, I'm a newbie around, so I kindly ask - please excuse.

I've seen some of you guys running other stuff like KeyHuntCuda, VanitySearch, Kangaroos and so on. I'm looking into BitCrack here. But I'm no Scientist, at most, a drunk sadistic alchemist... exploring around.

Here's the point where I have my doubts:

- BitCrack seems to have been correctly compiled, and runs;
- System has one GPU - RTX 4080 Super OC with 16 GB;
- OS Ubuntu Server 22.04, kernel 5.15.0-161-generic;
- System is periodically updated;
- Spent my time (dunno if it was enough though) trying to fine-tune blocks -b threads -t and points -p;

I've investigated a quote from a gentleman from around, where he states the following:
"
on a single 4090 i get ~3600 with -b 8192 -t 64 -p 368 (running more than 1 command on the same system u get out of memory error.)
on multiple 4090's on the same system best is -b 4096 -t 128 -p 256  and you get around ~3300 depending on card.. power..cooling etc.
using nvidia-smi you can see temps,fans and most important to check if you have a power cap.. 400-500W cap is ok under 400W you will see a drop in keyspeed.
"

- 4090's have 128 SM's, so instead of trying the block number as above,
- Multiplied my 80 SM's by the same number as the threads... that didn't work so well - 5120 blocks with 64 threads doesn't seem to reflect improvement, with a variety of point numbers.
- Locked on 64 threads, tried different block multipliers, and performed a reasonable amount of tests to conclude that odd is better, and above the half of the prior value. Either 3760 or 3920.
- Points, it looks like she likes being busy...

- My question - why do I keep seeing another guy running with speed values around ~3150 Mkey/s with the same GPU, while I only get ~2305 Mkey/s?
- Is there a secret formula

Thank you for your time, and will look for feedback

[realnewuser]

Hello, first and foremost, I'm a newbie around, so I kindly ask - please excuse.

I've seen some of you guys running other stuff like KeyHuntCuda, VanitySearch, Kangaroos and so on. I'm looking into BitCrack here. But I'm no Scientist, at most, a drunk sadistic alchemist... exploring around.

Here's the point where I have my doubts:

- BitCrack seems to have been correctly compiled, and runs;
- System has one GPU - RTX 4080 Super OC with 16 GB;
- OS Ubuntu Server 22.04, kernel 5.15.0-161-generic;
- System is periodically updated;
- Spent my time (dunno if it was enough though) trying to fine-tune blocks -b threads -t and points -p;

I've investigated a quote from a gentleman from around, where he states the following:
"
on a single 4090 i get ~3600 with -b 8192 -t 64 -p 368 (running more than 1 command on the same system u get out of memory error.)
on multiple 4090's on the same system best is -b 4096 -t 128 -p 256  and you get around ~3300 depending on card.. power..cooling etc.
using nvidia-smi you can see temps,fans and most important to check if you have a power cap.. 400-500W cap is ok under 400W you will see a drop in keyspeed.
"

- 4090's have 128 SM's, so instead of trying the block number as above,
- Multiplied my 80 SM's by the same number as the threads... that didn't work so well - 5120 blocks with 64 threads doesn't seem to reflect improvement, with a variety of point numbers.
- Locked on 64 threads, tried different block multipliers, and performed a reasonable amount of tests to conclude that odd is better, and above the half of the prior value. Either 3760 or 3920.
- Points, it looks like she likes being busy...

- My question - why do I keep seeing another guy running with speed values around ~3150 Mkey/s with the same GPU, while I only get ~2305 Mkey/s?
- Is there a secret formula

Thank you for your time, and will look for feedback

You should avoid using BitCrack! 2305 Mkey/s on an RTX 4080 is awful. You're leaving so much of your beautiful GPU's performance on the table. I'm using FixedPaul's modification of VanitySearch (https://github.com/FixedPaul/VanitySearch-Bitcrack) and getting a stable 3600 Mkey/s on an RTX 4070 Ti Super while running it at minimum voltage. Also, make sure you're using MSI Afterburner—it's an incredibly helpful tool

[SovereignScott]

Hey there,

Thank you for the feedback.

So I'm guessing here that you recommend Windows. I see that I can run it on Linux according to the Github page. Have you tried it?

KeyHunt Cuda is apparently known for missing some keys... occasionally. Experienced any how the same?

I'm thinking of giving it a try. I'll look deeper into it. Though seems a bit hard to find details about it.

[realnewuser]

Hey there,

Thank you for the feedback.

So I'm guessing here that you recommend Windows. I see that I can run it on Linux according to the Github page. Have you tried it?

KeyHunt Cuda is apparently known for missing some keys... occasionally. Experienced any how the same?

I'm thinking of giving it a try. I'll look deeper into it. Though seems a bit hard to find details about it.

No, I haven't tried compiling it under Linux myself. But given that it's based on the classic VanitySearch, I'd expect it to work without any issues, and it might even be faster.

Regarding the missed keys, yes, I've run into that as well. My workaround is to extend the search range by an extra 1 billion keys. From what I can tell, it happens because CUDA splits the search into millions of threads, and it seems to lose one key per thread.

[SovereignScott]

Greetings,

I guess you were right... almost doubled the work done in the same time

Current task START time: Sat Nov 15 19:39:37 2025
GPU: GPU #0 NVIDIA GeForce RTX 4080 SUPER (80x128 cores) Grid(8192x256)
Starting keys set in 0.61 seconds 
4206.1 MK/s - 105 BKeys - 2^36.61 [0.04%] - RUN: 00:00:25.0|END: 18:34:11.3 - Found: 0 

Versus 2305 MK/s from Bitcarck...

we'll see

thanks

[Vertex_VTX]

I built the first REAL interoperable blockchain — no bridges, no wrapped tokens, no custodians, and quantum-secure. AMA

I’ve been building something that I honestly thought was impossible until it actually worked.

I created a new blockchain prototype that does something NONE of the major chains (Ethereum, Polkadot, Cosmos, Solana, ZetaChain, Chainlink CCIP, LayerZero) have ever achieved:

✔ REAL interoperability — not the marketing version.

Not bridges.
Not wrapped tokens.
Not multisig.
Not oracles.
Not a 3rd party locking funds.
Not “cross-chain messages”.
Not “omnichain synthetic assets”.

🔥 My blockchain directly validates the native cryptographic signature of:

Bitcoin (ECDSA secp256k1 + UTXO)

Ethereum (EIP-155 signatures)

Polygon / Optimism / Arbitrum / BSC (all EVM)

Solana (Ed25519)

Avalanche C-Chain

Fantom

Base

All inside its own chain — natively, cryptographically, without trusting anyone else.

This means:

A user can prove ownership of funds on ANY blockchain purely by providing its native signature.

No bridging.
No mint/burn.
No wrapper tokens.
No custody.
No oracles.

Just pure cryptographic truth.
Exactly how blockchains are supposed to work.

On top of that, the chain is quantum-secure (NIST PQC):

ML-DSA (Dilithium)

ML-KEM (Kyber)

SPHINCS+

Hybrid QRS-2/3 signature redundancy

Why does this matter?

Because for the first time, a blockchain can be a universal verifier for ALL other blockchains.

Something no one — and I mean NO ONE — solved yet.

If you're curious, skeptical, or want to challenge the idea, AMA.
I have logs, proofs, and test transactions showing it working in real time.