Matrix Solvers

Solving a system of linear equations is a common numerical technique applied in many fields including fluid dynamics, thermal analysis, mechanical simulations, and economics. As simulations and models increase in complexity, organizations require high performance software to meet their growing computational needs. Several widely available optimized versions of BLAS and LAPACK libraries have been written to take advantage of CPU architectures. Recently, graphics processing units (GPUs) have shown potential to offer substantial performance gains when solving data-intensive calculations.

Acceleware provides software solutions to leverage GPU and other high performance architectures to solve linear equations, offering significant performance improvements over optimized CPU libraries.

What does Acceleware Provide?

Acceleware offers solutions to solve your intensive numerical computations:

Software Solutions – Core software that is optimized for GPU architectures to provide the maximum performance improvement for your application.
Professional Services – Acceleware can improve the performance of unique algorithms that use custom linear algebraic methods.

Acceleware Software Solutions – Linear Solver Technology

Acceleware provides software solutions that reduce the time to solve large linear systems.  Acceleware’s software solutions support several standard algorithms:

Iterative Solvers

Direct Solvers

*Single and double precision support is available for the solvers above.

Acceleware Professional Services – Customizing IP

Acceleware can customize our existing solver solutions to solve problems unique to your industry.  Some calculations require special solvers to reach a solution based on the nature of the problem.  Acceleware has expertise in adapting and tuning algorithms to achieve optimal performance for a given architecture. Contact Acceleware to find out how fast your applications can go!

Preconditioners

Many simulations require the application of a preconditioner to improve the convergence rate of the iterative solver.  Acceleware’s linear algebra libraries contain several preconditioners such as AMG, ILU, polynomial, and sparse approximate inverse methods.  Please contact Acceleware for more information about your specific solver requirements, as the design of preconditioners tends to be problem specific.

What are target applications?

Solvers are chosen based on the specific proprieties of the matrix.  Some typical applications are listed below:

Why Choose Acceleware

Acceleware offers your organization several advantages versus building the solution yourself:

• De-risking hardware platforms and drivers
  • Acceleware future proofs your applications by testing and debugging hardware platforms and determining the most stable solution for your application.
• Expertise to get best performance out of HW
  • Acceleware’s expertise in parallel programming allows us to tune your algorithm for optimal performance on parallel architectures.
• Ability to handle different HW interfaces (e.g., CUDA, OpenCL)
  • Acceleware manages the different programming interfaces allowing your application to connect to Acceleware’s library independent of the underlying hardware.
• Multi-GPU support for larger problems
  • Acceleware has had a successful track record in solving large problems by using more than one GPU in a system.
• Clustered GPU solutions
  • Acceleware can solve your largest problems by leveraging our MPI cluster capable libraries with GPUs to support your application.