Less than 1% of the code changes. Everything the code knows about your business stays where it is; only the run time changes.
A legacy library is years of business logic that nobody wants to re-derive. AADC leaves it in place and compiles it. No rewrite and no multi-year programme: an integration, then faster code.
Common challenges with legacy quant systems
AADC takes a fundamentally different approach to legacy modernisation. Instead of rewriting your codebase or migrating to new platforms, AADC works with your existing code. Through sophisticated JIT compilation and automatic algorithmic differentiation, AADC accelerates your legacy analytics while preserving every line of battle-tested business logic. The result: 6-100x faster execution and complete Greeks with less than 1% code changes.
Without a rewrite
Minimal integration effort, maximum impact
Keep institutional knowledge intact
Greeks faster than original pricing
A Tier-2 European bank fixed the performance of its front-office risk infrastructure with MatLogica AADC. They achieved 15-20x speedups compared to manual adjoint differentiation, reduced overnight batch from 8+ hours to 2 hours, and eliminated 70% of their modelling codebase while unlocking new revenue streams.
See how AADC accelerates existing SIMM implementations without rewrites
AADC C++ computes ISDA SIMM sensitivities at 34,000 evaluations/sec on CPU, 47× faster than GPU pathwise differentiation on NVIDIA H100. GPU brute-force fails to converge at scale due to noisy finite-difference gradients.
AADC C++ computes SIMM trade allocation gradients at 33,580 evals/sec, 25× faster than GPU. Optimises a 50K trade portfolio in 19 ms with 1-2 iterations to converge.
Use AADC to develop C++/Python models faster, without optimisation concerns. Get production-grade performance and automatic AAD from day one. Focus on models, not infrastructure.
Integrate AADC with open-source frameworks like QuantLib and ORE to get community validation plus proprietary-grade performance. No vendor licensing fees, automatic AAD for all models.
A workload that runs 6 to 1000x faster needs a fraction of the instances for a fraction of the time: 50 to 99% off the bill. Only compiled kernels go to the cloud; the source stays home.
Schedule a technical assessment to understand how AADC can accelerate your existing codebase.
By continuing to use this site you accept our use of cookies and acknowledge that all content, technology descriptions, and methodologies are protected by copyright and applicable patents. See our Privacy Notice.
