How to speed up the yuv->rgb16 convertion process?

Dear Everyone:

       I am developing a project using xvid. The video is 720x576x25fps,  VGA adapter

support only 16bit color depth.

       Since the video card (ct69030 ) used in my project  does not support overlay, 

and only support 16 bit  color  depth ,I have to change the output of xvid from I420 

to RGB565. However the decode speed slows down heavily!  

       If using I420 mode, the decode speed is about  80 fps.

       If changing to RGB565, the speed is only about 30 fps.

       So,

      1) I wanna know that who has the most fast X86 assemble programme  to 

convert YUV to RGB565???
     
       2) Do you have any sugestion for my project?

                THansk very much!!!