It seems like C2 outperforms GMS2 by quite a bit on this benchmark. Which makes me wonder why my own game performs significantly better in gms than construct. The game logic is structured about the same, and the gpu load should be more or less identical. Hmm...
People often accuse Construct/HTML5/browsers of having poor performance, but when it comes to actual measurements, it actually performs excellently.
I think it's quite possible that if you reimplemented your whole game exactly the same way in a different engine, it would still be slow. If you use a huge amount of intensive logic for example, that is still going to have to be run no matter the technology you use, unless you optimise it. I'd guess since events and script are so different, you actually used a lot less logic in GMS, so it's not really a fair test.