FICO has announced FICO Xpress 9.8, a new release of its Xpress Optimization suite that integrates NVIDIA CUDA-X libraries to accelerate optimization workloads. The company says Xpress 9.8 includes a GPU-accelerated implementation of the hybrid gradient algorithm, delivering up to 50x speedups on very large optimization problems.
FICO says the hybrid gradient algorithm targets extremely large, dense problems with tens or hundreds of millions of non-zeros in the problem matrix, and that it can deliver faster solutions due to lower memory overhead. In this release, GPU acceleration for the hybrid gradient algorithm is a beta feature, and Xpress 9.8 supports any NVIDIA GPU with NVIDIA CUDA Compute Capability 7.5 or later.
“The ecosystem has quickly embraced GPU solvers, and by leveraging NVIDIA CUDA-X libraries, FICO Xpress 9.8 is helping customers achieve breakthroughs in speed and scale, transforming operational efficiency and decision making,” said Alex Fender, Director of Decision Optimization at NVIDIA.
Beyond the hybrid gradient update, FICO reports performance improvements across other solver components in Xpress 9.8. The mixed-integer programming engine is now 14% faster overall than last year and 24% faster on models that require at least 100 seconds, while the global solver is now 68% faster overall and 5.3x faster for models that take at least 100 seconds.
FICO positions Xpress Optimization for deploying analytic models into end-user applications using optimization engines for linear, mixed integer, and nonlinear problems, including scenario analysis and what-if comparisons. FICO also offers a complimentary 60-day trial license of FICO Xpress 9.8.
Source: FICO
