Best performing implementations for deployment
Key metrics for production decision-making
| Metric | AADC Python | AADC C++ | JAX | Julia |
|---|---|---|---|---|
| Greeks Time (10Ã100K) | 0.7s | 1.0s | 8.4s | 15.5s |
| AAD Overhead | +23% | +54% | +152% | +297% |
| Multi-threading | Native | Native | XLA parallel | @threads |
| Memory Efficiency | Good | Excellent | Moderate | Good |
| Integration Effort | Low | Low | High (rewrite) | High (new lang) |
| License | Commercial | Commercial | Apache 2.0 | MIT |
Performance measured on 2x Intel Xeon Platinum 8280L with 8 threads. See Hardware Specs for details.
All benchmarks executed on enterprise-grade server hardware
| CPU | 2x Intel Xeon Platinum 8280L @ 2.70GHz |
| Cores | 56 physical (28 per socket), 112 threads |
| Architecture | x86_64, Cascade Lake |
| L3 Cache | 77 MiB (38.5 MiB per socket) |
| RAM | 283 GB DDR4 |
| OS | Linux kernel 6.1.0-13-amd64 (Debian) |
| Model | Asian Option Monte Carlo |
| Dynamics | Geometric Brownian Motion (GBM) |
| Timesteps | 252 (daily over 1 year) |
| Greeks | Delta, Rho, Vega (3 sensitivities) |
| Threads | 8 (configurable) |
| SIMD | AVX2 (4 doubles/instruction) |
| GCC | 12.2.0 (Debian) |
| Clang | 14.0.6 (Debian) |
| Python | 3.11.2 |
| NumPy | 1.26.x |
| AADC | 2.0.0 |