The Power of Two Libraries in One
X-ScalePETSc
Merging the bidirectional exchange of information between the PETSc library* and the MVAPICH2 library* to optimize the interprocessor communication within PETSc. The result is X-ScalePETSc that gives your teams the tools to iterate faster, improving workload performance to save time and cost.
Based on PETSc 3.20.4 and MVAPICH2 2.3.7. Applicable to MATMPIAIJ matrix type and its subclasses
Our user-friendly approach starts at installation with install and execution in one command.
Comprehensive support packages available, with more configurations continually being developed.
Up to 13% MatMult kernal-level benefit and 10% application-level benefit for small problem sizes, enabling your teams to find solutions faster, saving you time and cost.
Optimized small to medium message intra-node CPU-CPU communication for MatMult (the PETSc matrix-vector multiplication kernel widely used in its iterative solvers and preconditioners) to enhance the performance of your large-scale data challenges.
Performance Case Studies
Enabling Teams To Iterate Faster
Questions on X-ScalePETSc?
Our team of experts are available to answer your questions, show you a demo and help you determine how X-ScalePETSc can accelerate your most complex business challenges
