Beyond Speed: Understanding the True Meaning of Solver Performance

Whether you’re managing supply chains, optimizing financial portfolios, or improving manufacturing processes, the efficiency and robustness of your mathematical optimization solver will significantly impact your outcomes.  

At Gurobi, we’re constantly working to enhance our solver’s performance with each release—and Gurobi 11.0 is no exception.  

This post explores why solver performance matters and how the latest enhancements in Gurobi 11.0 can benefit users. 

Understanding Performance Metrics

Solver performance is critical because it directly influences solver runtime. Depending on your use case, this can refer to the time it takes to: 

• Find better, feasible solutions: Often, finding any feasible solution quickly is essential, especially in real-time applications. The time it takes to improve upon these solutions is equally important. 

• Prove optimality: Beyond simply finding a solution, proving that it is indeed the best possible solution can be computationally intensive. Efficient solvers reduce this time significantly. 

• Reach a certain solution quality: Sometimes, the goal isn’t just to find an optimal solution, but to reach a solution that meets a specific quality threshold within a limited time.

However, performance isn’t just about speed; it also encompasses robustness.  

A robust solver consistently performs well across different scenarios; its performance does not vary drastically from one instance to another, ensuring predictability and reliability. 

Modeling (API) performance is another important consideration. Efficient integration and interaction with the solver through APIs can greatly enhance the overall user experience and application performance.  

Beyond this, model performance also encompasses model construction and modification performance, or the ability to create and update model components at high speed. 

The Value of Better Performance

To determine what enhanced performance could mean for you, ask yourself the following questions: 

• What is the immediate business advantage of faster solve times in your solution? 

• What would happen if you were able to consider more optimal outcomes in the same timeframe? 

• How could your solution benefit from increased model complexity or accuracy? 

Your answers will be unique to your organization, but generally speaking, leveraging the best solving performance possible leads to better, faster decisions—an invaluable competitive advantage. 

Improved performance also allows for the inclusion of more complex and accurate models, which can capture a higher degree of granularity and lead to better, more realistic solutions. 

Gurobi: Pushing the Boundaries of Performance  

Gurobi is proud to produce the world’s most powerful solver with not only the fastest solve times, but a robust performance that consistently produces optimal solutions. 

The Gurobi R&D team has spent thousands of hours developing performance enhancements that Gurobi 11.0 users can leverage immediately, including: 

• Algorithmic Improvements: The latest version incorporates numerous algorithmic enhancements that automatically improve performance across various problem types. 

• New Techniques: One notable addition is the global nonlinear solver, which is enabled through a parameter setting. 

• MILP Performance: Mixed Integer Linear Programming (MILP) performance continues to improve with every release. Gurobi algorithms are now 80 times faster than they were in release 1. 

• Nonconvex MIQCP: Nonconvex Mixed Integer Quadratically Constrained Programming (MIQCP) has seen remarkable improvements as well, now 88 times faster since release 9. 

These performance enhancements mean that users can solve more complex problems more quickly, leading to better and faster decision-making. Moreover, users can achieve even higher levels of accuracy through the native handling of nonlinear model components. 

To learn more about these enhancements and how they can benefit your organization, check out the full list of features in Gurobi 11.0.