Linux vs. Windows: My computer stopped booting, so I replaced the motherboard (and only the motherboard). The Chipset changed from VIA KT-400 to VIA KT-600 (wow, big change. The driver for either chipset from VIA is the same!). This is what each OS did:

Linux: ''Your system configuration has changed. Do you wish to update Y/N?'' I press Y and enter. It boots up into X, no problem. After 2 keystokes, done.

Windows: For every device in my system: ''New hardware found.'' I press Next three times to get through all the wizard screens, yet to have to click ''Finish.'' Some devices even asked for the CD. This took 20+ mouseclicks, and opening the CD tray 3 or 4 times. That took about 5 minutes (not counting the time it took for me to look around my room to find the actual CD''s). After all that, it wanted to reboot.


With regards to performance: These are the specs of one of my graphics projects:
Windows .exe program in windows: 27 FPS
The same Windows .exe program running under WINE in Linux: 29 FPS

WTF!? In this case Linux faking Windows was slightly faster than Windows itself!

There's no reason why WINE might not be faster. Remember, it's not an emulation, it's a re-implementation.

Yeah it's probably not typical, but I see no reason why in some cases WINE can't be faster. Windows uses official opengl and linux uses Mesa. The speed of my code would run the same under either OS since once I set up the window, dc, and rendering context I make no calls to the windows API except for window event handling and file i/o. The only window events of importance to my program is close, mouse, keyboard, and resize, all rarely used, and the bottleneck of file access is the mechanical speed of the disc drive. I use vertex arrays, so there are few calls to GL, so it all comes down to Mesa vs OpenGL, and how well each interfaces to their OS and video driver. Under Immediate Mode, WINE would probably be slower than Windows since each gl* call would be translated from stdcall to unix calling convention, and then passed to Mesa.