A high performance C++ BLS library. This library features:
- High speed assembly implementation for x86_64
- G1/G2 Arithmetic & Pairing
- Aggregate Signatures
- EIP-2537
- EIP-2333
- Zero dependencies
- WASM build
Reference Libraries:
- zkcrypto/bls12_381 (Rust)
- zkcrypto/bellman (Rust)
- go-ethereum/bls12381 (Go)
- ConsenSys/bls12-381 (Go)
- Chia-Network/bls-signatures (C++)
- Relic-Toolkit (C)
mkdir build
cd build
cmake ..
make -j
ctest
./bench/eth_bench # eth benchmarks
./bench/chia_bench # chia benchmarks
The following numbers were generated running the above benchmarks (and their equivalents of Ethereum's and Chia's libraries) on an Intel(R) Core(TM) i7-8550U CPU @ 1.80GHz.
Ethereum's Go implementation of bls12381 vs. this library. Two different compilations are benchmarked:
go/cpp- Native go/cpp implementation is benchmarkedasm- The assembly implementation is benchmarked
Ethereum's library (go): This library (cpp): Ethereum's library (asm): This library (asm):
G1 Addition G1 Addition G1 Addition G1 Addition
886 ns/op 875 ns/op 743 ns/op 747 ns/op
G1 Multiplication G1 Multiplication G1 Multiplication G1 Multiplication
338003 ns/op 336085 ns/op 281708 ns/op 281852 ns/op
G2 Addition G2 Addition G2 Addition G2 Addition
2714 ns/op 2451 ns/op 1851 ns/op 1848 ns/op
G2 Multiplication G2 Multiplication G2 Multiplication G2 Multiplication
1075886 ns/op 1011230 ns/op 821738 ns/op 752900 ns/op
Pairing Pairing Pairing Pairing
1372026 ns/op 1231320 ns/op 1022350 ns/op 982783 ns/op
Chia Network's bls-signatures library vs. this library. Four different compilations are benchmarked:
cpu-ext-ct, asm- CPU features ADX and BMI2 are enabled at compile-time, the x86_64 assembly implementation is usedcpu-ext-rt, asm- CPU features ADX and BMI2 are enabled at run-time, the x86_64 assembly implementation is usedno cpu-ext, asm- CPU features ADX and BMI2 are disabled, the x86_64 assembly implementation is usedno cpu-ext, cpp- CPU features ADX and BMI2 are disabled, the C++ implementation is used
Chia Network's library: This library (cpu-ext-ct, asm): This library (cpu-ext-rt, asm): This library (no cpu-ext, asm): This library (no cpu-ext, cpp):
Signing Signing Signing Signing Signing
Total: 5000 runs in 5447 ms Total: 5000 runs in 4535 ms Total: 5000 runs in 4726 ms Total: 5000 runs in 6160 ms Total: 5000 runs in 8344 ms
Avg: 1.0894 ms Avg: 0.907 ms Avg: 0.9452 ms Avg: 1.232 ms Avg: 1.6688 ms
Verification Verification Verification Verification Verification
Total: 10000 runs in 32285 ms Total: 10000 runs in 19259 ms Total: 10000 runs in 19727 ms Total: 10000 runs in 24561 ms Total: 10000 runs in 33982 ms
Avg: 3.2285 ms Avg: 1.9259 ms Avg: 1.9727 ms Avg: 2.4561 ms Avg: 3.3982 ms
Public key validation Public key validation Public key validation Public key validation Public key validation
Total: 100000 runs in 26464 ms Total: 100000 runs in 2066 ms Total: 100000 runs in 2092 ms Total: 100000 runs in 3071 ms Total: 100000 runs in 4146 ms
Avg: 0.26464 ms Avg: 0.02066 ms Avg: 0.02092 ms Avg: 0.03071 ms Avg: 0.04146 ms
Signature validation Signature validation Signature validation Signature validation Signature validation
Total: 100000 runs in 28818 ms Total: 100000 runs in 12474 ms Total: 100000 runs in 12665 ms Total: 100000 runs in 17622 ms Total: 100000 runs in 24255 ms
Avg: 0.28818 ms Avg: 0.12474 ms Avg: 0.12665 ms Avg: 0.17622 ms Avg: 0.24255 ms
Aggregation Aggregation Aggregation Aggregation Aggregation
Total: 100000 runs in 287 ms Total: 100000 runs in 224 ms Total: 100000 runs in 226 ms Total: 100000 runs in 299 ms Total: 100000 runs in 410 ms
Avg: 0.00287 ms Avg: 0.00224 ms Avg: 0.00226 ms Avg: 0.00299 ms Avg: 0.0041 ms
Batch verification Batch verification Batch verification Batch verification Batch verification
Total: 100000 runs in 138664 ms Total: 100000 runs in 91959 ms Total: 100000 runs in 95714 ms Total: 100000 runs in 121560 ms Total: 100000 runs in 164145 ms
Avg: 1.38664 ms Avg: 0.91959 ms Avg: 0.95714 ms Avg: 1.2156 ms Avg: 1.64145 ms
PopScheme Aggregation PopScheme Aggregation PopScheme Aggregation PopScheme Aggregation PopScheme Aggregation
Total: 5000 runs in 14 ms Total: 5000 runs in 11 ms Total: 5000 runs in 11 ms Total: 5000 runs in 14 ms Total: 5000 runs in 21 ms
Avg: 0.0028 ms Avg: 0.0022 ms Avg: 0.0022 ms Avg: 0.0028 ms Avg: 0.0042 ms
PopScheme Proofs verification PopScheme Proofs verification PopScheme Proofs verification PopScheme Proofs verification PopScheme Proofs verification
Total: 5000 runs in 16567 ms Total: 5000 runs in 10315 ms Total: 5000 runs in 10531 ms Total: 5000 runs in 13400 ms Total: 5000 runs in 18593 ms
Avg: 3.3134 ms Avg: 2.0630 ms Avg: 2.1062 ms Avg: 2.6800 ms Avg: 3.7186 ms
PopScheme verification PopScheme verification PopScheme verification PopScheme verification PopScheme verification
Total: 5000 runs in 10 ms Total: 5000 runs in 5 ms Total: 5000 runs in 6 ms Total: 5000 runs in 7 ms Total: 5000 runs in 10 ms
Avg: 0.0020 ms Avg: 0.0010 ms Avg: 0.0012 ms Avg: 0.0014 ms Avg: 0.0020 ms
make clean
- Modular Multiplication using AVX2 section 11.16.2