Secp256k1 math is based on the excellent work from JeanLucPons/VanitySearch, and FixedPaul/VanitySearch-Bitcrack with major CUDA-specific modifications.
Special thanks to Jean-Luc Pons for his foundational contributions to the cryptographic community.
A distributed pool that performs brute-force key search for Satoshi puzzles on NVIDIA GPUs. The system has three components:
- Server: Assigns 40-bit work units (“segments”) randomly across a 70-bit search space with strict de-duplication. Tracks issued vs. completed segments, total pool throughput, and per-client speed; validates client result reports.
- Client: Connects to the server, receives a segment, computes it, submits the result, then requests the next segment.
- Telegram bot: Provides real-time pool statistics: total processed segments, currently in-flight work, per-client stats, and aggregate pool speed.
If a private key is found, it is submitted to the server for verification and payout:
- 1 BTC to the pool operator / software author.
- 1 BTC to the finder.
The remaining 5.1 BTC distributed pro-rata to contributors by completed segments (minimum qualification: 100 segments).
Credentials for pool (Telegram bot): @Satoshis_Quest_auth_bot
Telegram bot link: 70 bits pool
This is neccesary to acces to the pool!
To get started with 70 bit pool, clone the repository and type make
For totaly clean system (big thanks for dev_nullish):
apt update;
apt-get install -y joe;
apt-get install -y zip;
apt-get install -y screen;
apt-get install -y curl libcurl4;
apt-get install build-essential;
apt-get install -y gcc;
apt-get install -y make;
apt install cuda-toolkit;
git clone https://github.com/Dookoo2/Distibuted-CUDACyclone
make- --grid: very usefull parameter. Example --grid 512,512 - first 512 - number of points each thread will process in one batch (Points batch size)., second 512 - number of threads in one group (Threads per batch).
- --slices: batch per thread for one kernel launch.
- --ip: server ip address.
- --port server port for communication with a client.
- --worker: your user-id at the pool (MUST BE AT LEAT 6 CHARACTERS)!
Server ip address: 146.103.96.126
Server tcp port: 15763
Simple command to start with 70 bits pool.
./CUDACyclone --grid 1024,256 (try to use different grid for a better performance) --slices 8 --id YOUR ID FROM BOT!.txtIf you want, you could test your GPU with a different --grid parameter.
For top-tier GPU use --grid 1024,512 or -grid 512,512 for a fastest work.
For mid-tier GPU try to use --grid 256,256 or --grid 512,256.
RTX5090
Worker 1: mkeys=7914.915
Worker 2: mkeys=8586.059
Worker 3: mkeys=9007.229
It depends due to power limit and power consumption of GPURTX4070 TI Super
Worker 1: mkeys=3244.539RTX4060
Worker 1: mkeys=1254.091V1.3: Major UPDATE! Final version of a client! Speed up to 15% (4070 TI Super from 2.9 to 3.3 Gkeys/s
V1.2: Major UPDATE! Fix the bug with key skipping, adding security updates for proof of computing ranges
V1.1: Update of TG-bot
V1.0: Release