Linear solvers are key components in any software platform for scientific and engineering computing. The solution of large and sparse linear systems lies at the core of physics-driven numerical simulations relying on partial differential equations (PDEs) and often represents a significant bottleneck in data-driven procedures, such as scientific machine learning. In this paper, we present an efficient implementation of the preconditioned s-step Conjugate Gradient (CG) method, originally proposed by Chronopoulos and Gear in 1989, for large clusters of Nvidia GPU-accelerated computing nodes. The method, often referred to as communication-reduced or communication-avoiding CG, reduces global synchronizations and data communication steps compared to the standard approach, enhancing strong and weak scalability on parallel computers. Our main contribution is the design of a parallel solver that fully exploits the aggregation of low-granularity operations inherent to the s-step CG method to leverage the high throughput of GPU accelerators. Additionally, it applies overlap between data communication and computation in the multi-GPU sparse matrix-vector product. Experiments on classic benchmark datasets, derived from the discretization of the Poisson PDE, demonstrate the potential of the method.
communication-reduced algorithms
GPUs
linear solvers
s-step preconditioned Krylov methods
We present a solution based on a suitable combination of heuristics and parallel processing techniques for finding the best allocation of the financial assets of a pension fund, taking into account all the specific rules of the fund. We compare the values of an objective function computed with respect to a large set (thousands) of possible scenarios for the evolution of the Net Asset Value (NAV) of the share of each asset class in which the financial capital of the fund is invested. Our approach does not depend neither on the model used for the evolution of the NAVs nor on the objective function. In particular, it does not require any linearization or similar approximations of the problem. Although we applied it to a situation in which the number of possible asset classes is limited to few units (six in the specific case), the same approach can be followed also in other cases by grouping asset classes according to their features.
