MatLogica | QuantLib XVA Demo: 350x Faster with AADC

QuantLib XVA Demo: 350x Faster with AADC

Practical Example: MatLogica Supercharging QuantLib XVA Pricing

Watch this demo showing how MatLogica AADC achieved multithreading and a 350x speed-up for XVA pricing using QuantLib and ORE. Portfolio of swaps and swaptions: 20 hours reduced to 4 minutes with just 1 day of integration effort.

MatLogica AADC for QuantLib XVA Acceleration

This practical demonstration shows how MatLogica AADC was integrated with QuantLib and Open Risk Engine (ORE) to achieve dramatic performance improvements for XVA (Credit Valuation Adjustment, Debit Valuation Adjustment, Funding Valuation Adjustment) pricing of interest rate derivatives. Integration and Results: The MatLogica quant team spent just 1 day integrating QuantLib and ORE with AADC. As a result, the library was enhanced with automatic optimizations, native CPU vectorization (AVX2/AVX512), and safe multi-threading capabilities. The demonstration achieved a 350x performance improvement on XVA pricing for a portfolio of 5 swaps and swaptions. Real-world scalability: When scaled to real-life portfolios of 1,000 swaps and swaptions (typical for banks), the XVA calculation would take just under 4 minutes instead of 20 hours—a transformation from overnight batch processing to near real-time intraday pricing. Technical achievements: The integration provides automatic code optimizations without manual effort, native CPU vectorization leveraging AVX2/AVX512 instructions, multi-threading made safe automatically even if original QuantLib code isn't thread-safe, while preserving QuantLib's mathematical accuracy and pricing models. Developer access: MatLogica offers a free demo version of the AADC-supercharged QuantLib library for developers to test and evaluate. This enables hands-on experience with the performance improvements before full deployment. Applications: Ideal for XVA pricing (CVA, DVA, FVA), interest rate derivatives portfolios, swap and swaption valuation, counterparty credit risk calculations, and any QuantLib-based pricing that requires performance optimization for production use.
Video demonstration of 350x faster XVA pricing with MatLogica AADC and QuantLib

How to Speed Up QuantLib for XVA Calculations?

Our quant team spent just 1 day to integrate QuantLib and ORE (Open Risk Engine) with MatLogica AADC. As a result, the library was enhanced with automatic optimizations, native CPU vectorization (AVX2/AVX512), and safe multi-threading. We observe a 350x performance improvement on XVA pricing for a portfolio of 5 swaps and swaptions.

If scaled to real-life examples with portfolios of 1,000 swaps and swaptions (typical for investment banks), the XVA execution would take just under 4 minutes - instead of 20 hours. This transforms overnight batch processing into near real-time intraday XVA pricing capabilities.

Key Performance Metrics:

  • 350x faster XVA pricing with AADC
  • 1 day integration for QuantLib + ORE
  • 4 minutes vs 20 hours for 1,000 trade portfolio
  • Automatic optimizations - no manual code changes
  • Safe multi-threading - even if original code isn't thread-safe
  • Native vectorization - AVX2/AVX512 support