Introduction

Modern applications in mechanical design, in architecture, and in many other areas of engineering require a powerful and reliable geometric constraints solver, capable to handle large quantities of data and to resolve huge systems of geometric constraints equations.

Autodesk® uses D-cubed solver for implementing 2D constraints both in AutoCAD® and in Autodesk® Inventor®. However, even this production-proven solver, which provides geometry foundation for most of the leading CAD applications, has its significant limitations.

Cheetah Solver opens the way for parametric modeling beyond these limitations. It can work with geometric models that contains hundreds times more geometric and dimensional constraints, and resolve them thousands times faster than you can do with the constraints solvers of the present CAD applications.

CAD market requires continuous improvement in performance, reliability and productivity. CAD users appreciate the ability to modify geometric models and to regenerate them as fast as possible. Rapid design as well as rapid editing of complex geometric models are among the most important challenges of parametric CAD. Automated rework of models leads to lower engineering costs and fewer errors. The more easily and cheaply those changes can be made, the more design variations and refinements designer can explore.

Constraints solvers, integrated in applications of the leading CAD vendors, are able to resolve models containing only some hundreds of constraints, but even these models, which are trivial for our Cheetah Solver, are often problematic for all commercial CAD programs – user often need to wait many seconds (and even many minutes) to regenerate such a model.

Let us run AutoCAD and look at the “Performance” tab of the Task Manager (where you should set the CPU graph to “Logical processors”). Let us start dragging some complicated parametric model. You can notice (see Fig.1 on the next page) that AutoCAD uses mainly only one logical processor, while other logical processors practically do nothing. Note that, for instance, Intel Core i7 processor has eight logical processors. Thus, during intensive work of constraints solver, total utilization of CPU in AutoCAD is only about 15%.

The reason of such a weak exploration of a multi-core CPU is that algorithms used in the present solvers do not use parallel computing. The situation is quite different with Cheetah Solver – our solver can use the full power of modern multi-core CPUs.

Fig. 1. Working with AutoCAD solver. Total utilization of CPU is 16%. Only one core is working.

 

Fig. 2. Working with Cheetah Solver. Total utilization of CPU is 66%. All cores are working.

Marketing our Cheetah Solver as a third-party plugin for AutoCAD, we believe that most AutoCAD users will benefit from greater performance and reliability of our solver comparing with the built-in solver of AutoCAD and solvers of other parametric CAD applications.

Fig. 3. Switching On/Off Cheetah solver.

However, when you start using Cheetah Solver plugin for AutoCAD, it does not necessary means that you should say farewell to the built-in solver of AutoCAD. You can easily switch back and forth from one solver to another (see Fig. 3 above). For some projects, it might be useful (for instance, for the reasons of compatibility) to continue using native solver of AutoCAD. In many scenarios, it will be useful to modify a model in Cheetah mode and then, may be, convert it into native AutoCAD model.

And, certainly, we do believe that many AutoCAD users will prefer to work in the Cheetah mode and not to return to the native AutoCAD mode.

Leave a reply