A high-performance, CPU-optimized tool for computing public keys on the secp256k1 elliptic curve. It includes features for searching both compressed and uncompressed public keys, with customizable search parameters. Discuss on bitcointalk.org.
- 🍏 Fixed 256-bit modular arithmetic
- 🔄 Group inversion for point addition operations
- 🍇 Precomputed tables for point multiplication
- 🔍 Search for compressed and uncompressed public keys (hash160)
- 🌟 Accelerated SHA-256 with SHA extension (both ARM and x86)
- 🚀 Accelerated RIPEMD-160 using SIMD (AVX2/NEON)
- 🎲 Random search within customizable bit ranges
- 🍎 Works seamlessly on macOS and Linux
- 🔧 Customizable search range and thread count for flexible usage
git clone https://github.com/vladkens/ecloop.git && cd ecloop
make build* On macOS, you may need to run xcode-select --install first.
By default, cc is used as the compiler. Using clang may produce faster code than gcc. You can explicitly specify the compiler for any make command using the CC parameter. For example: make add CC=clang.
Also, verify correctness with the following commands (some compiler versions may have issues with built-ins used in the code):
make add # should found 9 keys
make mul # should found 1080 keysUsage: ./ecloop <cmd> [-t <threads>] [-f <filepath>] [-a <addr_type>] [-r <range>]
Compute commands:
add - search in given range with batch addition
mul - search hex encoded private keys (from stdin)
rnd - search random range of bits in given range
Compute options:
-f <file> - filter file to search (list of hashes or bloom fitler)
-o <file> - output file to write found keys (default: stdout)
-t <threads> - number of threads to run (default: 1)
-a <addr_type> - address type to search: c - addr33, u - addr65 (default: c)
-r <range> - search range in hex format (example: 8000:ffff, default all)
-q - quiet mode (no output to stdout; -o required)
-endo - use endomorphism (default: false)
Other commands:
blf-gen - create bloom filter from list of hex-encoded hash160
bench - run benchmark of internal functions
bench-gtable - run benchmark of ecc multiplication (with different table size)
For a quick start with Bitcoin Puzzles, there are preconfigured make commands. Simply run them, and ecloop will start searching for the puzzle in random mode. For example, use make 71 for puzzle 71, make 72 for puzzle 72, and so on. If you are lucky and find a key, the result will be saved in the found_N.txt file. Shortcuts available for: 28, 32, 36, 71, 73, 74, 75, 76, 77, 78, 79.
./ecloop add -f data/btc-puzzles-hash -t 4 -r 800000:ffffff -o /tmp/found.txt-fis a filter file with hash160 values to search for. It can be a list of hex-encoded hashes (one per line) or a Bloom filter (must have a.blfextension).-tsets the number of threads (e.g., 4).rdefines the start:end of the search range.-ospecifies the file where found keys will be saved (if not provided,stdoutwill be used).- No
-aoption is provided, so onlyc(compressed) hash160 values will be checked.
cat privkeys.txt | ./ecloop mul -f data/btc-puzzles.blf -a cu -t 4cat privkeys.txt– the source of HEX-encoded private keys to search (can be a file or a generator program).-f– a Bloom filter containing hash160 values to search for (may produce false positives but has a much smaller size; for example, all BTC addresses ever used take up ~6 GB).-a– specifies which type of hash160 to search:cfor compressed,ufor uncompressed,cuto check both.-t– sets the number of threads (e.g., 4).
ecloop can also take a raw word list and automatically hash each word with SHA-256. Use the -raw flag to enable this:
cat wordlist.txt | ./ecloop mul -f data/btc-puzzles.blf -a cu -t 4 -rawThe rnd command allows you to search random bit ranges within a specified range (by default, the entire curve space). This mode is useful for exploring random subsets of the keyspace.
./ecloop rnd -f data/btc-puzzles-hash -o ./found.txt -a cu -qThis command performs a random search across the entire keyspace, checking both compressed and uncompressed addresses (-a cu), and saves the results to found.txt. Quiet mode is enabled (-q).
A random bit offset will be used, with a 32-bit range per iteration (4.2M keys).
./ecloop rnd -f data/btc-puzzles-hash -d 128:32This command searches a 32-bit range with a 128-bit offset (-d 128:32). It will execute a search with a random base key on each large iteration. Bits from offset to offset + size will be dynamic. For example:
iter1: d33abfe4b9152c08 7176d067XXXXXXXX 27d4419e6969a205 4d1deb10e4929621
iter2: 1b354d3094405c2f bf8f5c15XXXXXXXX 46804248255476e9 800f26edd71ad0d7X represents dynamic bits; other bits are randomly generated during large iterations.
Note: You can also combine random search with -r param for shorter ranges.
cat data/btc-pu
A4D0
zzles-hash | ./ecloop blf-gen -n 1024 -o /tmp/test.blfcatreads the list of hex-encoded hash160 values from a file.-nspecifies the number of entries for the Bloom filter (the number of hashes).-odefines the output file where the filter will be written (the.blfextension is required).
The Bloom filter uses p = 0.000001 (1 in 1,000,000 false positives). You can adjust this option by modifying n. See the Bloom Filter Calculator.
A list of all addresses can be found here or use bcloop to make dump from Bitcoin Node.
Created Bloom filter then can be used in ecloop as a filter:
./ecloop add -f /tmp/test.blf -t 4 -r 8000:ffffffNote: Bloom filter works with all search commands (add, mul, rnd).
Get the performance of different functions for a single thread:
./ecloop benchShould print output like this:
_ec_jacobi_add1: 6.70M it/s ~ 0.90s
_ec_jacobi_add2: 5.44M it/s ~ 1.10s
_ec_jacobi_dbl1: 5.47M it/s ~ 1.10s
_ec_jacobi_dbl2: 7.81M it/s ~ 0.77s
ec_jacobi_mul: 0.02M it/s ~ 0.55s
ec_gtable_mul: 0.32M it/s ~ 1.54s
ec_affine_add: 0.31M it/s ~ 1.63s
ec_affine_dbl: 0.31M it/s ~ 1.63s
_fe_modinv_binpow: 0.20M it/s ~ 0.50s
_fe_modinv_addchn: 0.31M it/s ~ 0.32s
addr33: 4.84M it/s ~ 1.03s
addr65: 4.27M it/s ~ 1.17sNote: This benchmark is run on a MacBook Pro M2.
Here are the steps I followed to run ecloop on Windows:
- Open PowerShell
- Run
wsl --install - Restart Windows.
- Run
wsl --install Ubuntu(this command hung when I tried it, so I continued in a new tab) - Run
wsl -d Ubuntu - Run:
sudo apt update && sudo apt install -y build-essential git clang - Run
cd ~ && git clone https://github.com/vladkens/ecloop.git && cd ecloop - Run
make build
If no errors appear, ecloop has been compiled successfully and is ready to use. For example, you can run a benchmark with: ./ecloop bench.
Tests were done on an Intel N100.
> time ./keyhunt -m rmd160 -f ../ecloop/data/btc-puzzles-hash -r 8000:fffffff -t 1 -n 16777216
3m53s ~ 1.15 MKeys/s
> time ./ecloop add -f data/btc-puzzles-hash -t 1 -r 8000:fffffff
1m06s ~ 4.09 MKeys/s> time ./keyhunt -m rmd160 -f ../ecloop/data/btc-puzzles-hash -r 8000:fffffff -t 4 -n 16777216
1m31s ~ 2.95 MKeys/s
> time ./ecloop add -f data/btc-puzzles-hash -t 4 -r 8000:fffffff
0m25s ~ 10.73 MKeys/sThis project was created to explore the mathematics behind elliptic curves in cryptocurrencies. The functionality for searching Bitcoin Puzzles was added as a real-world use case.
If you find this project useful, consider supporting its development:
- BTC:
bc1q4c3mxpm50awx9qaprx54k5x3t5m9ex658yzk4j - ETH:
0x4DF8E04C5AC0b06fb9938581D8a1732D5A78703E - SOL:
4r3CeYxvwJa1btZudmLpHeu2yzeudRw4UTqMUZScD63j - XMR:
85yjYN1sU3sgGFEkdqLKKxdiQwzQjqyhT74m5j4xwmKqHYfensRMjJrB1HvE9H6R6G5wG938KDpkJLum6GQd5q5yTTk8uhj - Buy Me a Coffee
Thank you for your support!