problem with Automatic parallelization in GCC4.5

[zhouxu(NUDT)]

Hi,
    I have successfully build GCC4.5 on my Fedora12. And I want to
test the automatic parallelization feature of GCC4.5. I write a system
program and compile it with flags:
-O2 -fno-loop-strip-mine -floop-interchange -fgraphite-identity
-ftree-parallelize-loops=4 -floop-parallelize-all
-fdump-tree-graphite-all
But I got the following error:
block.c:1:0: sorry, unimplemented: Graphite loop optimizations cannot be used
I am confused. It seems the error is related to the flag
"-floop-parallelize-all". Does anyone knows what the problem is?
Thanks.

-- 
zhouxu