3. Solvers

Release 2026 R1 includes the following improvements to the solution process:

3.1. Resource Prediction Feature

The new PREA command introduces resource prediction, a feature that estimates computational resource requirements before solving. This helps identify the optimal hardware environment for simulations.

By analyzing the current load step, analysis type, solver type, and core count, PREA provides estimates for:

  • Solution elapsed time

  • Total memory usage

  • Total disk space usage

  • GPU recommendation

These predictions support distributed-memory parallel (DMP) solutions and assist with performance and resource allocation. While predictions offer valuable guidance, they do not guarantee solution success or solver convergence.

3.2. PCG Block Lanczos Eigensolver

The preconditioned conjugate gradient (PCG) Block Lanczos eigensolver is now officially released.

This eigensolver is a variant of the Block Lanczos method designed for large symmetric eigenvalue problems in modal analyses. It uses the same automated shift strategy as the standard Block Lanczos method but adds PCG iterative solver technology alongside the existing sparse direct solver.

To use this eigensolver, combine the Block Lanczos method (MODOPT,LANB) with the PCG solver (EQSLV,PCG). This combination automatically activates the new eigensolver, which improves performance and memory efficiency.

For more information, see the MODOPT and EQSLV commands.

3.3. PCG Solver Enhancements

The PCG solver requires less memory than in previous releases to solve models when:

  • Running on higher core counts (more than 64 cores).

  • Constraint equations and/or coupling equations exist in the model.