PyCudaBTCMiner

A Python-based Bitcoin mining application with GPU acceleration using NVIDIA CUDA technology. Even with the most powerful GPUs, this application cannot match the speed of dedicated ASIC mining devices. Therefore, it employs a random sampling strategy, hoping to find valid nonces faster than sequential iteration methods.

Prerequisites

1. NVIDIA GPU with CUDA Support

   • GPU must be compatible with CUDA toolkit

2. NVIDIA CUDA Toolkit (Version 10 or higher)

   • Download from: https://developer.nvidia.com/cuda-downloads
   • Ensure CUDA Toolkit is installed and available in system PATH

3. Python 3.x

   • Required to run mining scripts

File Structure

CUDA Files (GPU Kernels)

• liblooper.cu - CUDA kernel source code for sequential nonce scanning from 0 to 4,294,967,295 (0x00000000 - 0xFFFFFFFF)
• rr.cu - CUDA kernel source code for random nonce scanning

Build Scripts

• build.bat - Compilation script for Windows (generates .dll files)
• build.sh - Compilation script for Linux (generates .so files)

Python Modules

• looper.py - Python module serving as interface between miner.py and CUDA kernel (liblooper)
• rrnonce.py - Python module serving as interface between random_miner.py and CUDA kernel (rr)

Mining Applications

• miner.py - Mining application using random extranonce2 and full nonce scan (0-4294967295)
• random_miner.py - Mining application using random extranonce2 and random nonce

Testing Scripts

• test.py - Testing script for full scan and GPU hashrate benchmark (using block 933,995)
• test_1_block.py - Testing script using Bitcoin's first block header (Genesis Block)

Documentation Files

• working_script.png - Screenshot of the application running
• Readme.md - This documentation file

Usage Instructions

1. Compile CUDA Kernels

For Windows:

build.bat

For Linux:

chmod +x build.sh
./build.sh

The compilation process will generate:

• Windows: rr.dll and liblooper.dll
• Linux: rr.so and liblooper.so

2. Testing Kernel and Benchmark

Before starting mining, run tests to ensure CUDA kernels are working properly:

python3 test.py

This script will:

• Test CUDA kernel functionality
• Measure your GPU's hashrate
• Display time required to complete full scan (0-4294967295)

3. Mining Configuration

Edit miner.py or random_miner.py according to your needs:

config = {
    "pool_address": "public-pool.io",  # Mining pool address
    "pool_port": 3333,                 # Mining pool port
    "user_name": "YOUR_BITCOIN_ADDRESS", # Replace with your Bitcoin address
    "password": "x",                    # Password (usually "x" or "password")
    "min_diff": 1,                      # Minimum difficulty (minimum is 1)
    "poll_sleep": 0.05,
    "reconnect_backoff": 5.0,
}

Important Notes:

• Replace user_name with your own Bitcoin address
• min_diff minimum is 1 (kernel only outputs hash if difficulty ≥ 1)

4. Start Mining

Method A: Random Extranonce2 + Full Nonce Scan

Recommended for GPUs with hashrate > 4,294 MH/s

python3 miner.py

This method will:

• Use random extranonce2
• Perform full nonce scan from 0 to 4,294,967,295
• Suitable for very powerful GPUs

Method B: Random Extranonce2 + Random Nonce

Recommended for all GPUs

python3 random_miner.py

This method will:

• Use random extranonce2
• Perform random nonce sampling (196,608,000 nonces per extranonce2)
• More efficient for medium-speed GPUs

5. Adjusting Random Nonce Sampling (Optional)

If you're using random_miner.py and want to increase the number of nonces tried per extranonce2:

1. Open random_miner.py file
2. change batch_number in config:

config = {
    "pool_address": "127.0.0.1",
    "pool_port": 3333,
    "user_name": "bc1qngzehzs73x2p5k7r7pa7na69ej89p40qxnrh60",
    "password": "x",
    "min_diff": 1,
    "batch_number": 42,   # 1.048.576 nonce per batch x 42 = 44.040.192 sample nonce per extranonce2
    "poll_sleep": 0.05,
    "reconnect_backoff": 5.0,
    "max_extranonce2": 0xFFFFFFFF
}

3. If you want to change how many nonce launch per batch edit rr.cu`line 134, 135 int threadsPerBlock = 256; int blocksPerGrid = 4096; nonce launch per batch = 256 * 4096 = 1.048.576
4. Recompile kernel by running build.bat or build.sh

Understanding Mining Concepts

Difficulty Target

• Mining pool sets target difficulty for each share
• Higher difficulty means harder to find valid shares
• min_diff in configuration is the minimum difficulty the miner will search for

Extranonce2

• Part of coinbase transaction that can be modified by miner
• Changing extranonce2 produces different merkle root
• Each extranonce2 provides 4,294,967,296 possible nonces

Random vs Sequential Strategy

• Sequential (ASIC): Tries nonces 0, 1, 2, 3, ... in order
• Random (PyCudaBTCMiner): Tries random nonces hoping to find faster

Monitoring and Output

When mining is running, you'll see output like:

[+] New Job ID detected! Switching to job: 6978d6ba00000709 Pool difficulty : 1.0
  [✅] 02:16:09 Found Share!
       Job ID: 6978d6ba00000709
       EN2   : 755c8940
       Nonce : d0ee7abf
       Hash  : 000000000fb1abc2d054395e4aaec12fb8c4d2d4a962056e4c656b6708b8c6
       Diff  : 260.99 (Target from pool: 1.00)
       Hashrate : 4294.96 MH/s

Troubleshooting

Error: "liblooper.so not found" or "rr.dll not found"

• Ensure you've run the build script (build.bat or build.sh)
• Verify .so or .dll files exist in the same directory as Python scripts

Error: "NVCC not found"

• CUDA Toolkit is not installed or not added to PATH
• Download and install from: https://developer.nvidia.com/cuda-downloads

GPU not detected

• Ensure latest NVIDIA drivers are installed
• Run nvidia-smi to check GPU status

Very low hashrate

• Your GPU may not be powerful enough for Bitcoin mining
• Try using random_miner.py for better efficiency

Support

If you encounter issues:

1. Ensure all prerequisites are met
2. Run test.py to verify CUDA kernel
3. Check error logs for details
4. Create an issue on GitHub repository (if available)

---

Developer Notes:

• I use Linux for development
• Windows script (build.bat) has not been thoroughly tested
• If you encounter issues on Windows, please report them

License: Feel free to use for educational and experimental purposes. But if you succeed in getting block rewards, please donate to me, my btc address bc1qngzehzs73x2p5k7r7pa7na69ej89p40qxnrh60