Anyone knows of good methods to optimize my VGA emulation while keeping it accurate as it is now? It currently converts horizontal and vertical timings into a horizontal and vertical lookup tables, combining them for every horizontal and vertical pixel. It then processes sequencer data (pixel data) to attribute data, finally converting it to DAC data (RGB values) writing them to the display surface. When a frame has been finished, it will be resized if needed(by the GPU rendering routines) to fit the screen if needed. In Windows this usually is set to direct plot, meaning a simple copy of rendered pixels 1:1 to the display.

My source code:
https://bitbucket.org/superfury/x86emu.git

The VGA emulation is at /hardware/vga