Background
In aircraft design and design optimization, engineers take into account static and dynamic loads and modes of vibration. Similarly, engineers need to study the deformation of the car’s frame under both normal and extreme loads in order to better protect the car passengers. Biomechanical engineers need to evaluate the stress generated within a stented arterial blood vessel. It is important to study the risk of a product or process before actually manufacturing the product or implementing the process. In order to evaluate or study all the events mentioning above, engineers often build a physical prototype. The problems with building physical prototypes involve cost and time. Sometimes it is difficult or impossible to fine-tune prototypes to closely match the real objects or events. For example, how do you build a prototype of the bridge damaged by the San Francisco earthquake in 1989 in order to study its structure so that newer bridges can be built that are less prone to collapse during an earthquake?
Simulation was created to answer engineering design questions. Simulation helps reduce costs related to building prototypes and physical testing. It is possible to study many computational prototypes rapidly to understand the behavior of specific materials in order to make a better choice of materials for certain products. Simulation not only helps reduce costs related to building prototypes and testing, it also eliminates the cost involved in correcting post-manufacturing design errors. Simulation helps optimize product design, predict performance, and examine a product’s behavior.
This article will discuss software simulation from ANSYS1 called ANSYS Mechanical v17.02 and how ANSYS Mechanical software is able to gain performance boost when running on a system equipped with Intel® Xeon® processor E5 v4.
ANSYS Mechanical
From the ANSYS website: “ANSYS Mechanical software is a comprehensive finite element analysis (FEA)3 tool for structural analysis, including linear, nonlinear, dynamic, hydrodynamic and explicit studies. It provides a complete set of elements behavior, material models and equation solvers for a wide range of mechanical design problems. In addition, ANSYS Mechanical Enterprise offers thermal analysis and coupled-physics capabilities involving acoustic, piezoelectric, thermal–structural and thermo-electric analysis.”
FEA is a computerized method of simulating the behavior of engineering structures and components under a variety of conditions. It works by dividing a model into a number of finite elements. Each element is represented by a set of equations. By integrating all the element’s equations, the whole object can be mathematically modeled.
In general, FEA consists of three steps: preprocessing, analysis, and postprocessing. In the preprocessing step, the object (the spoon in this case) is divided into multiple elements to create a mesh. The next step is to feed the information creating from the preprocessing step to the analysis step to create a mathematical model of that object. This model consists of a set of linear or nonlinear algebraic equations. At the end of the analysis step, the solution of this equation set is solved and the results are passed to the next step, postprocessing. At the postprocessing step, the results are displayed in a form that is easy to understand and evaluate.
Picture 1 shows the head of the spoon. The spoon has been discretized as shown in Picture 2.
Picture 1: Object to be analyzed.
Picture 2: The object has been discretized.
Benefit of Running ANSYS Mechanical 17 on Intel® Xeon® Processor E5 v4
ANSYS Mechanical utilizes FEA to construct a set of equations to be solved, and among the solvers available in the solution phase is the sparse direct solver4. The sparse direct solver may be called hundreds or thousands of times in the solution phase in order to solve increasingly accurate linear approximations to the overall nonlinear set of equations. Any improvement to the performance of the sparse direct solver will boost the overall performance of ANSYS Mechanical.
In ANSYS Mechanical, the sparse direct solver makes frequent calls to the function DGEMM (Double-Precision General Matrix Multiplication)5. ANSYS Mechanical makes use of a version of the DGEMM function available in the Intel® Math Kernel Library (Intel® MKL)6. Intel MKL and functions, including DGEMM, are optimized to use Intel® Advanced Vector Extensions 2 7, and so run optimally on the Intel® Xeon® processor E5-2600 v4 product family. By incorporating Intel MKL function calls, ANSYS Mechanical can update the sparse direct solver for the latest Intel® architecture by simply updating the Intel MKL distributed with their product.
Performance test procedure
To prove that ANSYS Mechanical v17.0 performs better on newer Intel® processor-based systems, we performed tests on two platforms. One system was equipped with Intel® Xeon® processor E5-2697 v3 and the other with Intel® Xeon® processor E5-2697 v4. The latest generation Intel Xeon processor E5-2600 v4 product family features a higher core count, faster memory, larger cache, and other improvements.
Test configurations
System equipped with Intel Xeon processor E5-2697 v4
- System: Preproduction
- Processors: Intel Xeon processor E5-2697 v4 @2.3 GHz
- Cache: 45 MB
- Cores: 18
- Memory: 128 GB DDR4-2133MT/s
System equipped with Intel Xeon processor E5-2697 v3
- System: Preproduction
- Processors: Intel Xeon processor E5-2697 v3 @2.6 GHz
- Cores: 14
- Cache: 35 MB
- Memory: 128 GB DDR4-2400 MT/s
Operating system: Red Hat Enterprise Linux* 6.4
Software:
- Intel MKL
Application: V17SP-5
Test results
Figure 1: Comparison between the Intel® Xeon® processor E5-2697 v3 and the Intel® Xeon® processor E5-2697 v4.
Figure 1 shows the results on a system equipped with the Intel Xeon processor E5-2697 v3 and on a system equipped with the Intel Xeon processor E5-2697 v4. On the Intel Xeon processor E5-2697 v4, a performance improvement of 70 percent was achieved due to more cores, faster memory, and core improvements.
Note: Software and workloads used in performance tests may have been optimized for performance only on Intel® microprocessors. Performance tests, such as SYSmark* and MobileMark*, are measured using specific computer systems, components, software, operations and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products. For more information go to http://www.intel.com/performance
Conclusion
Simulation helps build prototypes faster and reduce cost involved by replacing physical prototypes and physical testing with computational prototypes and testing. The simulation engine in ANSYS Mechanical gains performance by running on a system equipped with the Intel Xeon processor E5 v4 compared to running on a system equipped with the Intel Xeon processor E5 v3 due to improved hardware architecture, more cores per socket, larger cache size, and faster DDR4 memory.