Cost Optimizing Your Rescale Job

Rescale provides a suite of software with flexible on-demand licensing aimed at optimizing computational simulations by allowing users to pay based on usage. The software is designed to encourage running simulations on more cores for faster results, with pricing models that include proportional, flat rate, capped, and discounted categories. The blog post examines how different pricing models, along with the scalability of the software, influence the overall cost-effectiveness of running simulations on multiple cores. It highlights the importance of understanding both hardware and software pricing to minimize total costs, which are calculated as the product of hourly prices and simulation duration. The discussion includes an exploration of the relationship between compute and communication times in distributed simulations, emphasizing that larger models tend to scale better due to reduced relative communication overhead. By presenting equations and assumptions, the post aims to guide users in optimizing their simulations' cost efficiency, ultimately suggesting that a good scaling software combined with a flat rate or capped pricing model often proves more economical. Rescale's support team is available to assist in estimating the optimal number of cores for specific simulations, reinforcing the idea that minor deviations in core numbers generally do not significantly affect cost savings.