any slowness from a PiP swap is due to the wow graphical engine changing its directX buffers and internal rendering doodads (very technical )

there could also be an impact due to the sizes of the buffers within graphics memory.

v1.8g will make the large area swap out first.. hopefully clearing up some memory for the smaller area scaling up. but to be honest, with a 512M card rendering a 1600x1200x3 screen ... that's only about 5M of memory per screen.. unless its the various bitmaps used to wrap the surfaces.. which might be rendered once then shipped to the card (making more sense as i type it...). that rendering could take time and is prolly done cpu side then ship to grfx memory (then again, my directX knowledge is a bit rusty)

if anyone has any insight.. please enlighten