I know this needs more work. Part of the problem is that on many projects Shotcut is barely able to keep up with real time let alone play faster than real time. So, when Brian says “2X” what it does is render every other video frame. Unfortunately, when the engine is put into that mode, it no longer cares about real time, and simply feeds every other frame through the engine as fast as it can. That speed depends on a lot of things, but on one of my main systems a 280x720 ProRes with no filters plays a minute of footage in 2 seconds in that mode!
Also, now I understand why the sound is clicking/popping when scrubbing/forwarding. How do streaming websites speed up the audio so seamlessly at higher speeds? There must be an interpolation algorithm already in the public domain for that (and I bet it uses Fast Fourier Transforms!).