QML is commonly used on <1 GHz embedded devices. Equating QML to bad Electron examples doesn't make sense. (Although Electron/Chromium also have a large range depending on how they are used, in my experience they have a way harder time scaling down to smaller devices)

> QML is commonly used on <1 GHz embedded devices.

Yes, I know as an owner of Symbian 9.x device ;)

But, here I literally means that actual QML coding style stuck to requirement of modern GPU with OpenGL 2.x (GL ES 2.x) even OpenGL 3/4 support.

My laptop has integrated GPU mostly limited to OpenGL 1.x, so all those CSS/QSS animations, which depends on higher OpenGL features, just blows my CPU/GPU into flame.

> QML is commonly used on <1 GHz embedded devices.

1ghz embedded devices with semi-good GPUs*

e.g. the GPU in a Pi 3 blows some older intel GMA chips out of the water, and those are still quite present "out there". Even a 4ghz pentium 4 with such a GPU will feel slower to the user than a 800mhz Pi (which in itself, is not what I'd call enjoyable to use).

Was thinking more the various Vivante things like you get with the imx6 variants - but they also do support the needed OpenGL ES versions. Pis are weird outliers in various ways (and Pi 3 is >1Ghz).

Your 10yo GPU can blit and scroll pixels faster than your CPU, which can deal with layouts and other things in the meantime. Using GPU by itself is not slowing down your apps.

If the app is badly optimised it's going to be slow regardless of the rendering method.