Benchmarking is essential for making informed decisions about computational resources. When you launch a simulation, its performance and cost depend heavily on how well the available hardware — CPU cores, memory, and GPUs — is used. Without systematic testing, you risk paying for resources you don’t use efficiently or waiting longer than necessary for results.
A Benchmark lets you test different machine types and configurations to find the sweet spot: the setup that gives you the best performance for the lowest cost. Simply adding more cores doesn't always speed things up linearly. By running a Benchmark, you can see exactly how your simulation scales and avoid wasting money on hardware that doesn't provide a real benefit.
You should make benchmarking a regular part of your workflow. Consider it whenever you are:
The easiest way to start is by running a shorter, representative version of your full simulation across a few different machine configurations. After this short run, check the performance results in the Web Console to see how efficiently your simulation used each machine or configuration. Based on these insights, you can confidently choose the optimal configuration for your full-scale workload, knowing you're making the most of your time and budget.
In short, a quick benchmark today saves you from unnecessary spending and delays tomorrow.