The design of hardware tests can completely alter the perception of each components performance. From testing on different systems, different installations, and obviously using different benchmarking software the results can be interpreted in various ways. Many reviews are written with a single use in mind and the testing is all geared to getting the optimum results based on the intended audience.
The testing used for these articles will generally follow the exact same format.
- Testing is done on a clean system that is checked for any driver or other compatibility issues.
- For comparing individual components, the same system minus the component being tested will be used.
Real World Tests
- CAD Applications – Settings are saved specifically for testing in all applications. The scripts running the tests are designed to record the time and other information to complete typical daily operations. Each software uses a production model and rather than just opening and spinning true tasks are performed such as rebuilds, adding features/blocks, inserting components, view changes, generating dimensioned drawings, running simulations and generating toolpaths.
- Software List – Typically the following will be tested but the list is subject to change depending on the test. For example, when comparing graphics card performance software that does not utilize the GPU will sometimes not be tested.
- PhotoView 360
- SolidWorks Simulation
- SolidWorks Flow Simulation
- Autodesk Inventor
- Fusion 360
- Office Test – In addition to the usual testing of individual software the “Office Test” is running the SolidWorks and AutoCAD scripts while Outlook, Excel, and Internet Explorer are running operations every 10 seconds. While this is overboard compared to a normal daily operation it gives insight into multitasking.
Performance testing wouldn’t be complete without the normal benchmarking numbers. These are especially useful for testing quick changes such as overclocking, hard disk upgrades, software updates and other workstation changes.
- SPECviewperf – In testing engineering workstations SPECviewperf simply has to be on the list. It is designed specifically for measuring performance in professional applications.
- Cinebench – Another testing tool designed for professional applications is Cinebench. This tool provides a heavier testing of the CPU and GPU with it’s higher floating point precision.
PassMark – The PassMark testing is widely used to test the entire workstations performance and breaks down the individual performance sections. While it is not primarily designed for professional applications it gives additional information about the performance and allows simple baseline testing and comparisons.