I found out about this site from NASA Watch (hat tip to Keith Cowing). Check out the following URL:

http://www.sandroid.org/Apollo/index.html

Some very bright people at MIT with lots of free time on their hands have created an emulator for the Apollo Guidance Computer (AGC). If you live in the San Francisco bay area, you can see an actual (non-functioning) AGC at the Computer History Museum on Shoreline Blvd. in Mountain View (a terrific museum). At one time the AGC was the world's most spiffy computer.

I down loaded the Bzip2 tarball from the AGC website onto a Linux system. It was a straight forward configure/make operation to build the executable followed by "make install" as root to render the code useable (It won't work correctly until properly installed under /usr/local). After it is installed, you invoke the command "SimColossus249" to bring up a GUI that looks like the AGC's control panel as seen by the Apollo astronauts. The user interface is the same as what the astronauts actually used. The emulator drives the same assembly language used by the original AGC. As it stands, the AGC emulator is only partially complete. The computer science aspect is more-or-less finished but the really hard part of emulating the inertial measurement unit (IMU) and atmospheric entry does not exist. When I first found this thing, I was hoping the all the control logic for atmospheric entry was there and operational. The raw assembly code for atmospheric entry may actually be there but there is no emulated hardware for it to drive. However the real jackpot with this MIT website is the documents section. They have almost all of the declassified Apollo AGC software and documentation available on-line. This is a real gold mine! As soon as I saw it, I immediately downloaded everything. This information is simply not available anywhere (it's not in the NASA Ames library).

The Apollo Command Module (CM) flew a very sophisticated entry trajectory for lunar return that kept the astronauts under a peak decelleration of 6 gravities while not putting an excessive heat load on the heat shield. This involved lots of roll modulation at constant angle-of-attack using the Reaction Control System (RCS) that was controlled by the AGC. The Apollo CM trajectectory utilized a form of aerocapture where the vehicle plunged into the atmosphere in a "lift-down" orientation, then performed a snap roll to "lift-up" at minimum altitude to deliberately skipped back out of the atmosphere. The CM then re-entered a second time and deployed its parachute. During the whole process, the roll angle was being constantly adjusted to keep the G loading from getting too high. Getting the G loading right is really hard. I've been playing with Apollo CM simulations but haven't been able to get the G loading right (I'm always running over 9 Gs with a naive simulation flying at constant roll angle). The good news is I now have that information. The bad news is the information is buried inside approximately 500 Mbytes of extremely technical pdf documents.

Gary