Fast bitwise swap?

[jj2007]

Celeron M:

Code Select Expand

9 cycles, qword 256-byte-lut, 313 bytes
9 cycles, drizz stir-fry, 64 bytes
8 cycles, drizzNew stir-fry, 61 bytes

[drizz]

AMD Opteron(tm) Processor 148    (SSE3)

Code Select Expand

5 cycles, qword 256-byte-lut, 313 bytes
4 cycles, drizz stir-fry, 64 bytes
4 cycles, drizzNew stir-fry, 61 bytes

[WryBugz]

10 cycles, qword 256 byte lut
8 cycles, drizz stir fry
5 cycles drizznew stir fry

I lose!

So much for Intel Core I-7
Anyone got an xt I can have?