When it comes to comparing browser performance relying on most readily available benchmarks can cause confusion. For a good TV experience benchmarking the smoothness of the UI’s animations is key.
There are some very good rendering benchmarks such as MotionMark and OrangeMark that give an aggregate score based on testing a range of visually rich features including HTML, CSS and canvas. UIs tend to be written using a combination of HTML & CSS. Canvas is less used as it lacks the declarative nature of HTML meaning animations have to be handled programmatically which can lead to lower performance.
It’s therefore important to look beyond the headline scores for these benchmarks and focus only on the performance of the features that are needed. Fortunately, most benchmarks also provide further detail which can be used to compare browser performance for the target UI. Combining the performance data only for the features used will give a good indication of the smoothness of the animations on the final product.
When testing Flow, our new GPU enabled browser, we wanted to be sure that we were comparing it against the best possible implementations of the leading browser engines (Blink, Gecko & WebKit); this meant testing Chrome, Firefox & Safari on desktop hardware. Testing on target hardware is also vital as desktop machines and embedded systems often have very different configurations for processor connectivity and memory architecture and that can have a big impact on realisable performance.
At Ekioh we’ve been looking at a whole range of TV UIs and applications and using them to draw up a list of must have features. Over the coming months, we’ll be writing about these features and providing benchmarking data from popular third party benchmarks which highlights our performance against them.
A more general discussion on browser benchmarking can be found in our white paper “Investing in product success through careful browser selection”.
If you’d like to automatically receive our posts by email please join our mailing list.