What percentage of users really get WebGL and hardware acceleration?
The graphs below show the evolution over time of the percentage of users who really get WebGL and hardware acceleration.
The main reason why a user might not get these features is driver blacklisting. Another possible reason is insufficient hardware.
Table of contents
- WebGL is enabled by default on all platforms, but it's not used by most users, so the WebGL demographics are biased towards power users. It is the feature that's closest to depending only on having recent enough drivers.
- Accelerated Layers is used by default on all platforms except on Linux. In addition to requiring recent enough drivers, it has further requirements that rule out some old/low-end hardware, and a stricter driver blacklist.
- For Direct2D and DirectWrite, see the Direct3D 10 Accelerated Layers curve and discussion in the Windows 7 results.
- The vertical axis is the success rate, and the unit is percentage. That is, the ratio of people who actually get the feature over people who try to get it.
- The horizontal axis is time.
- Mouse over the graph to show the color code. WebGL is blue, Accelerated Layers is orange, Direct3D 10 Accelerated Layers is green.
General results over all operating systems
Mouse over to highlight individual values. Click and drag to zoom. Double-click to zoom back out. Change the number and hit enter to adjust the averaging period.
Windows results over all Windows versions
Windows 8 results
Windows 7 results
Windows Vista results
Windows XP results
Mac OS X results
Linux (GNU) results
Percentage of sessions where WebGL context creation was attempted
OS Market share among Firefox crashers
Number of crash reports each day
Crash reports contain information about which graphics features have been attempted, and which ones successfully initialized. This is true even if the crash itself has nothing to do with graphics features.
By analyzing these crash reports, we determine, among all crash reports sent on a given day, the number of crash reports reporting that a graphics feature had been attempted, and the number of crash reports where it had successfully initialized. The ratio between these two values is what is shown below as "success rate" percentage.
Obviously, getting this data from crash reports biases the results to a certain extent. Graphics features might be correlated with crashiness, for example, although crash-stats suggests that's not too marked.