Performance Testing: Overview
At BSN we break out testing into two parts Synthetic and Real-World. Each has an important part in the overall review process. With Synthetic testing you have an easily repeatable performance measure than anyone can use against the same hardware configuration. But Synthetic testing cannot hope to provide a completely accurate picture of performance. These testing suites do not have the capacity to take multiple factors into account. On the opposite side are the real-world tests. They are duplicates of how a system [or part] will perform in actual use. However as they take into account the dynamic nature of real-world use they are hard to duplicate. As such our testing will always be an average of three runs on each test. The results shown here will be that average. We feel this combination offers the best overall picture of performance and how well the product will perform for the consumer when they get it home and installed in a system.
Each of our synthetic tests cover a different aspect of system performance. Taken as a whole they provide a very broad overview of how well each board or product we have in the lab will perform for you once you get it home.
PCMark Vantage is a suite of tests that covers the most complete range of system task possible. It is true that it cannot hope to cover every possible activity but it does an excellent job of covering the majority and providing the end user with a good idea of how well a system can complete common tasks.
The X58A-UD9 did an excellent job of pushing through all of the tests in the PCMark Vantage suite. We are fairly confident that this is probably due to this suite of tests not being terribly HDD dependent.
HyperPi is an application that is capable of running multiple instances of SuperPi XS 1.3 on a system. It allows you to select the number of instances you wish to run as well as the length you want to calculate the number Pi out to [up to 32 million places]. For our testing we run one instance per core both physical and logical. On something like the i7 870 this is a total of eight 32M instances which puts a healthy strain on the CPU, memory and dive. It is also a good indication of how well the mainboard can handle the large amounts of data being passed back and forth.
As HyperPi is very memory and CPU dependent the X58A-UD9 was able to plow through the 12 instanced of SuperPi 32M running. This shows that the board can handle the overhead generated by that type of work load and keep on going.
Cinebench is a synthetic render test developed by Maxon. Maxon is the maker of Cinema 4D, one of the industry standards in 3D animation. This test covers CPU based rendering as well as how your GPU/system is able to handle OpenGL instructions. Cinebench is capable of testing rendering against a single CPU core as well as spreading the rendering task across all available cores. For our testing we used the R-11.5 64-bit version.
Although the margins here are not large the UD9 still came out ahead in both the CPU test and the OpenGL test. Again, as with most synthetics, the solid memory performance more than likely came into play here.
© 2009 - 2013 Bright Side Of News*, All rights reserved.