It looks like flash 8 could be *considerably* faster for complex vectors with gradients and alpha.
These two bitmap caching examples have been put together by some clever hacky dudes:
http://www.margaris.de/flash8/bitmap/bitmapcache.html
http://www.potapenko.com/tmp/test_8_flash.html
More examples of new features here:
http://osflash.org/doku.php?id=flash8
I'm now getting pretty excited....
