Graphics, Hardware, Software Programs

MWC14: Imagination Technologies Shows Off 3DMark Running OpenGL ES 3.0


Here at the Mobile World Congress in Barcelona, there has been a lot of talk about OpenGL ES 3.0 and OpenGL ES 3.0 benchmarks, the popular companies like Kishonti, Futuremark and Rightware are all showing their latest benchmarks with OpenGL ES 3.0. Some of them are showing it running on various hardware, but the truth is that right now the benchmarks are making all hardware grind to a halt. 

Imagination Technologies has been showing off Futuremark’s OpenGL ES 3.0 benchmark, which is actually a complete rewrite of their 3DMark Cloud Gate benchmark that we’ve already been using. They did this using Imagination Technologies’ PowerVR Series6 GPUs which can be found in a lot of devices today. Also, not forgetting that Apple is also using Imagination Technologies’ GPUs in their products which means that we will be able to see that from them soon-ish.

Not to leave others out, but we’ve also seen OpenGL ES 3.0 benchmarks running on Intel hardware to similarly difficult scenes and low frame rates. The only hardware that we haven’t seen here at MWC running a current-gen OpenGL ES 3.0 benchmark is actually Nvidia’s which would probably be because there isn’t any Tegra K1 hardware here at MWC for them to show us. It will be interesting to see how many companies release OpenGL ES 3.0 benchmarks and how many of those will get updated to the OpenGL ES 3.1 API that is expected to get announced at GDC next month. I would expect to see a comparison of various devices’ OpenGL ES 3.0 performance to come towards the second half of this year as more vendors’ hardware comes out and the benchmarks mature to reasonable levels.

Right now, the benchmark companies like Kishonti, Rightware and Futuremark are all indicating that their OpenGL ES 3.0 benchmarks are going to be ready for release within a few weeks if not sooner. This is in contrast to the game developers who are still struggling to release OpenGL ES 3.0 titles to make this API set and benchmarking a worthy exercise to begin with. After all, why benchmark something that isn’t actually getting any use?