Automatic Core Optimisation

Set the limit. Let the application optimise.

Specify the maximum number of CPU cores available to the modelling run. AERMOD One evaluates the modelling configuration and determines the appropriate number of cores to use within that limit.

  • User-defined maximum core limit
  • Automatic optimisation within that limit
  • No separate parallel workflow
  • Automatic execution strategy
How core selection works
1
User sets the maximum available cores
2
AERMOD One evaluates the modelling configuration
3
The run uses the appropriate number of cores without exceeding the selected limit
Example: if 16 cores are selected and 11 are appropriate, the application can use 11. If 8 are selected, it will not exceed 8.

Local by Design

Your data stays on your system.

Environmental modelling projects can contain sensitive information about facilities, emissions, stack parameters, building layouts, receptors and project assumptions. AERMOD One intentionally performs parallel computation locally rather than requiring the modelling workload to be sent to remote one-computing infrastructure.

POWERRUN · PERFORMANCE

Intelligent Parallel Processing.
Built In.

AERMOD One automatically determines how many local CPU cores to use for each modelling task—up to 16 cores—based on the configuration and computational requirements of the job.

No separate parallel-processing application. No per-core upgrade. No manual core optimisation. No one compute job. AERMOD One intelligently uses your local workstation to accelerate both AERMOD and AERMAP.

Intelligent local execution
Model configuration Analyzing
Selecting an efficient local execution strategy
↓
CPU
4
CPU
8
CPU
12
MAX
16
AERMOD One decides
The 16-core capability is an upper limit—not a setting the modeler needs to manage.
Built Into the Application

One application. One modelling workflow.

Parallel processing is integrated directly into AERMOD One. The same application handles model preparation and execution, without a separate parallel-processing executable or workflow.

One application

Model preparation and execution remain in AERMOD One.

No separate workflow

Parallel execution is part of the normal modelling process.

No per-core purchase

Parallel capability is included within the application.

INTEGRATED EXECUTION
AERMOD + AERMAP
Parallel local computation
Two Computational Bottlenecks

Accelerate more than just AERMOD.

Parallel processing is integrated into both AERMOD and AERMAP workflows, including demanding terrain processing associated with high-density receptor networks.

AERMOD

Accelerate dispersion calculations

Use intelligent local parallel execution for computationally demanding AERMOD modelling runs.

  • Large receptor grids
  • Long meteorological datasets
  • Multiple sources
  • Repeated scenarios
  • High-volume modelling studies
AERMAP

Accelerate terrain processing

High-density receptor networks and large terrain domains can make AERMAP processing a significant part of the modelling workflow.

  • Large terrain domains
  • High-resolution elevation data
  • Dense receptor networks
  • Complex terrain studies
  • Repeated terrain processing

Practical Performance

More Cores Aren't Always Better

Parallel workloads generally gain performance as cores are added, but the improvement becomes progressively smaller. Beyond an efficient operating point, additional cores can provide little benefit or reduce overall execution efficiency.

Parallel scaling
Performance
124816
Efficiency
124816
Conceptual illustration only. Relative performance can increase with additional cores while parallel efficiency declines as coordination, memory and non-parallel work become a larger part of total execution.

Practical Engineering

Up to 16 cores, with performance in mind.

Parallel performance does not scale linearly with the number of CPU cores. As more cores are added, the incremental performance benefit can diminish because of workload characteristics, memory bandwidth, I/O and other execution overheads.

AERMOD One supports up to 16 cores. The modeler sets the maximum number of cores available for a run, and the application determines the appropriate number to use within that limit.