김경종
4.2 KiB
type, title, complexity, domain, created, updated, address, aliases, tags, status, related, sources, source_refs
| type | title | complexity | domain | created | updated | address | aliases | tags | status | related | sources | source_refs | ||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| concept | Abaqus Resource and Parallel Execution | advanced | computational-mechanics | 2026-05-29 | 2026-05-29 | c-000077 |
|
|
current |
|
|
Abaqus Resource and Parallel Execution
Definition
Abaqus resource and parallel execution settings control how an Abaqus analysis uses memory, scratch disk space, CPU cores, MPI processes, threads, domains, and GPU acceleration.
How It Works
Environment settings can be defined at site, user, or job level and use Python syntax. Command-line options can override environment defaults. The guide highlights settings such as cpus, domains, parallel, run_mode, scratch, memory, standard_parallel, double_precision, and gpus.
Resource planning separates persistent output data from temporary scratch data. Disk space is needed for output files, memory is needed for performance-critical data, and scratch storage supports solver working data. Data check runs can estimate memory requirements before a full analysis.
Parallel execution differs across products and procedures. Abaqus/Standard can parallelize input preprocessing, direct sparse solution, iterative solution, and element operations. Abaqus/Explicit uses domain decomposition for most computation, while some preprocessing and packaging remain serial. User subroutines must be thread-safe when used in threaded execution.
Volume II adds procedure-specific performance levers: direct versus iterative linear solvers, explicit mass scaling, selective subcycling, matrix generation, substructuring, and co-simulation all change the computational cost profile of a model.
Why It Matters
Large finite element analyses often fail or become impractical because of resource configuration rather than formulation alone. Memory, scratch placement, parallel mode, solver choice, and user subroutine behavior can change runtime, scalability, and failure modes.
Connections
- Abaqus Job Execution Workflow supplies the command context where resource options are applied.
- Static Equilibrium Equation Solvers and Direct Time Integration Methods explain why implicit, explicit, direct, and iterative procedures stress resources differently.
- Finite Element Program Implementation provides the broader data-structure and solver context.
- Abaqus Explicit Analysis Efficiency Techniques covers explicit time-increment cost controls.
- Abaqus User Subroutines and Utility Routines captures extension-code memory and parallel-safety concerns.