Will someone please send me the ASM v2.01 and V1.5
im not sure which one came with my calculator. I got it late January this year
also, will someone please inform me of this HW 1 and HW2. I am confused on it

I am sorry to say this, but some experiments which I made gives to me a conclusion that the situation with AMS2.03/HW2 is worse than it looks like. Namely, it seems that HW2 calcs have built-in a hardware device which does not allow program counter to be located anywhere out of 8K long zone located at the top of the RAM. So, I want to say that 8K limit on HW2 is supported by hardware, not only by software. Breaking of 8K limit in Doors and UniOS is based on fact that ASM program can be executed anywhere in the RAM. It seems that this is not true on AMS2.03/HW2. I made some test programs which will prove or reject this hypothesis, and I need to send them to somebody which have real HW2 calc, so I expect that this hypothesis will be proven (or rejected) very soon (in next 2-3 days). If it is proven as true, I want to alarm whole TI community about this disasterous discovery.

There is still one chance. As it is possible to break 8K limit on HW2 calcs with AMS1.05, it seems that hardware limiter is turned on by software, somewhere in AMS2.03 (probably by writing to some I/O port). Now, it is a question whether this ugly feature may be TURNED OFF by software or not. If can, the problem is solvable very easily. If not, the situation is hopeless. If this is the case:

1) Forget kernels and libraries, and make pure "nostub" ASM programs which are smaller than 8K.
2) In this case the only way to make longer programs is dividing them into 8K chunks, which must share the same address space!

Btw, ASM programs which are shorter than 8K and does not use kernels and libraries works fine on AMS2.03/HW2. Such programs are, for example, my games creversi and cblaster.

I don't remember hearing about an 8k limit on ams 1.05...

*off topic question*

is it possible to make something that runs programs directly from the archive, instead of either unarchiving before running or just making a copy of the archived program into the RAM?

Sorry, error in phrasing: I want to say "As the 8K limit does not exist on AMS1.xx with HW2 etc etc"

Answer to the "off topic question": This is just what I would try next. Btw, the answer is surely NO if the program is self-modifing, or if it contains static data which can be modified, but if this is not the case, the answer is maybe YES. If so, it can give a way to solve the problem. This would be a next experiment...

so how would high scores be saved/games be saved? wouldn't it be easy just to have the game use a datafile (that wouldn't be archived)?

I have a HardWare version 2 calculator that came with 2.03, I downgraded to 1.05 and DoorsOS .93 runs but certain games crash (Tetris, Jezzball) but SMQ and Sobokan and Asteroids work fine

Tetris89 works perfectly on my TI-89 HW2 AMS 1.05 calculator with DoorsOS II 0.93 (except that there's no grayscale)...I think your problem is that you haven't overwritten the libraries contained within your calculator with the libraries that came with Tetris89. After installing DoorsOS, you should transfer all of the Tetris files along with their accompanying library files and overwrite the default libraries that came with DoorsOS. You also need a gray4lib as well for Tetris89 (it won't show grayscale but it is still required in order for you to run the application).

I tried using tetris89 on a friends calc (hw2) with doors .93 but it doesn't seem to work; it runs, but there's a load of crap all over the screen and runs incredibly slow

Why won't this work on my AMS 2.03 HW 2 (I think) TI-89?????????????????? THanks!