Dell M6500 review addendum: Vsync and Nvidia graphics performance

If you're willing to sacrifice some image quality, you can blow the roof off the Viewperf graphics benchmark

InfoWorld is a resource intended for use by managers and IT staff. As such, we test whole systems, rather than components, and we test those systems as configured by the vendor for the typical user. This approach ensures that you will have a similar experience if you purchase or use the same equipment -- without needing to rely on extraordinary or unusual measures to replicate our experience.

In the case of the Dell Precision M6500 mobile workstation, we used Dell's default setting for video, which includes the setting of vertical sync=on (in the driver for the system's Nvidia Quadro FX 3800M video card). Vertical sync, generally called vsync, is turned on by Dell because this setting provides the best possible image quality. However, it limits the video performance to running at the monitor's refresh rate. Turning this feature off can significantly improve graphics performance -- at the cost of slight image degradation. This degradation, however, is small enough that it was essentially invisible to us in tests, although a discriminating user with a good eye might see differences.

How much faster were the benchmarks with vsync turned off? The following table shows the results:

Nvidia Quadro FX 3800M graphics performance in the Dell Precision M6500

Benchmark Vsync on (from review) Vsync off
Viewperf 46.53 86.54
Cinebench / Speedup 2.94 x 3.00 x
Cinebench / Multiple CPU 11781 11915

Viewperf is the more comprehensive of the two benchmarks for testing graphics subsystem performance. The result in the right column, 86.54, is a huge number and better than most desktop workstations can muster. It shows a near doubling of performance when measured with vsync enabled. In actual practice, while the graphics will indeed render faster, they probably won't approach the 100 percent improvement the benchmark indicates. This is because much of the difference comes from large swings in two of the eight Viewperf viewsets.

The Cinebench benchmark measures how well the graphics subsystem leverages the processor for rendering an image. As can be seen, disabling vsync does not materially change this number. (The variance is well within the margin of difference between any two runs on the same machine.) If your graphics work consists of lots of high-quality static rendering work, it doesn't make much of a difference which setting you use. But for all other cases where graphics performance is important, turn vsync off.

This story, "Dell M6500 review addendum: Vsync and Nvidia graphics performance," was originally published at Follow the latest developments in mobile computing and hardware at