A LOT less janky compared to prior version. The windows timer fires anywhere from 8 to 20milliseconds so at this time the best I can do is compute the delta time from the last update to give a more accurate step. I have yet to figure out how to update at the refresh rate of the monitor consistently.
The WPF version is an experiment that I was hoping would fix the jankiness since it uses DirectX. Stats below are based on a 4.4Ghz i7-3770k, GTX1070, 1920x1200 resolution, and a 1ms windows timer resolution.
WPF Version:
- In maximized uses 90MB Memory + 25MB GPU Memory
- In fullscreen, uses 120MB Memory + 70MB GPU Memory
- Uses 10% less CPU during normal running (.44% vs .49%). Has to run hours (because this has a 500millisecond start up cost vs Winforms) until the WinForms version catches up to Total CPU time.
- Uses 2,000% (2 thousand %) more CPU in full screen. Needs investigation.
- Is much smoother than the WinForms version but not smooth like I expect it to be.
WinForms Version:
- In maximized, uses 28MB Memory + 0MB GPU Memory
- In fullscreen, uses 30MB Memory + 0MB GPU Memory
- Uses 10% more (0.49% vs 0.44) CPU during normal running
- CPU usage does not increase when running in full screen mode.
- Is much more janky than the WPF version