Trying out the different "wmode" params you can pass to flash as part of the tag.
"direct" and "gpu" are hardware accelerated, new w/ Flash 10, but come with a laundry list of caveats
and indeed, don't seem all that much faster than the software rasterizer. what's up with that.
here is an article titled "What does GPU acceleration mean?" by someone from Adobe, which talks a bit about each mode.

wmode = normal
wmode = transparent
wmode = opaque
wmode = direct
wmode = gpu       oxeFlash5.swf