MultiRun User Guide: Step-by-Step Optimization for Peak Performance
In modern engineering and simulation workflows, running a single scenario is rarely enough. To find the absolute best design, engineers must explore hundreds of parameter combinations. MultiRun is designed exactly for this purpose: automating parametric sweeps and parallel execution to find optimal configurations rapidly.
This guide provides a practical, step-by-step framework to configure, execute, and analyze your MultiRun setups for maximum computational efficiency. Scenario 1: Local Workstation Optimization
If you are running MultiRun directly on a local desktop or engineering workstation, your primary goal is maximizing CPU and RAM utilization without freezing your operating system. Step 1: Define Your Parameter Matrix
Open your configuration file (config.yaml or project settings).
List your independent variables (e.g., dimensions, velocity, voltage).
Define the sweep type: Discrete values, linear ranges, or logarithmic scales.
Example: Set velocity: range(10, 50, 5) and thickness: [2, 4, 6]. Step 2: Configure Local Resource Limits
Check your system’s physical core count (excluding hyper-threading threads for heavy simulations).
Set your parallel execution limit to N-1 or N-2 cores (where N is your total logical cores).
This reservation prevents your operating system and user interface from locking up during execution.
Ensure your total concurrent runs do not exceed your available system RAM. Step 3: Execute and Monitor
Launch the MultiRun sequence via the command line or GUI interface.
Monitor the active console log for thread utilization and initialization errors.
Watch for early memory warning signs to adjust pool sizes if necessary. Scenario 2: Cloud or HPC Cluster Scaling
When moving beyond a single workstation, MultiRun scales across High-Performance Computing (HPC) clusters or cloud infrastructure using job schedulers like Slurm or PBS. Step 1: Package Dependencies and Inputs
Containerize your simulation environment using Docker or Singularity.
Ensure all execution nodes have strict access to identical software versions.
Place all baseline mesh, geometry, or input decks in a high-speed shared network directory. Step 2: Set Up the Cluster Scheduler Script
Configure the MultiRun batch script to interface with your specific cluster manager.
Define your resource requests explicitly per simulation node:
#SBATCH –nodes=4 #SBATCH –ntasks-per-node=16 #SBATCH –time=12:00:00 Use code with caution.
Enable MultiRun’s internal load balancer to distribute individual parameter cases dynamically across allocated nodes as they become free. Step 3: Deploy and Aggregate Submit the master batch job to the cluster queue.
Track standard output logs (stdout) to verify that tasks are distributing evenly without idle time.
Utilize the automatic aggregation tool to merge split output files into a unified master database. Post-Run: Data Extraction and Performance Tuning Analyze Results Efficiently
Filter Failures: Immediately isolate and remove any simulation runs that exited with errors or failed convergence.
Generate Response Surfaces: Export the aggregated data to build 2D contour maps or 3D response surfaces to visualize optimal design pockets.
Identify Bottlenecks: Review the MultiRun execution log to see if specific parameter combinations took disproportionately longer to solve, signaling stability issues. Peak Performance Tips
Use Design of Experiments (DoE): Instead of a dense, brute-force grid sweep, use Latin Hypercube Sampling (LHS) to find optimal zones using 70% fewer total runs.
Disable Unneeded Outputs: Turn off high-resolution visual logging for intermediate iterations. Only write out essential scalar data to minimize slow disk writing operations (I/O bottlenecks).
To help tailor a more specific execution script or architecture setup for your team, tell me: What simulation engine or software is MultiRun driving?
What operating system and hardware environment (Local PC, AWS cloud, Slurm cluster) are you targetting? Saved time Comprehensive Inappropriate Not working
A copy of this chat, including the images and video, will be included with your feedback A copy of this chat will be included with your feedback
Your feedback will include a copy of this chat and the image from your search
Your feedback will include a copy of this chat, any links you shared, and the image from your search.
Thanks for letting us know
Google may use account and system data to understand your feedback and improve our services, subject to our Privacy Policy and Terms of Service. For legal issues, make a legal removal request.