So what about other architectures? How about a 1.6GHz Atom 330 processor. This doesn't have out of order execution and should more closely match ARM and MIPS type small machines. (Sorry, I can't get one of those booted from where I am now.)

  FOURCC                        23.3ns
  FOURCC-without-alignment-bug  27.6ns
  NGINX                         29.8ns
  STRCMP                        70.5ns  (glibc)

The difference between FOURCC and NGINX is vanishing here.

How about a 3GHz core-i7 840

  FOURCC                         4.8ns
  FOURCC-without-alignment-bug   5.6ns
  NGINX                          8.3ns
  STRCMP                        13.9ns  (freebsd)

Not a lot faster than a Core-2, but some. Of course it has twice the cores plus hyperthreading ready to go.

How about if you are cheap (like me) and got a 2.93GHz core i3 for your server:

  FOURCC                         5.3ns
  FOURCC-without-alignment-bug   5.8ns
  NGINX                          9.9ns
  STRCMP                        33.6ns  (glibc)

Not bad for the little guy. (Saved $100 on the cpu, plus didn't need to add a video card with some goofey driver, and its system wide low power state is almost as good at the Atom's.)