ok lets see the first ,,screenshot'' from the sysetm.
simplifyed C - half way beethwen C and binary code.
it seems from a 60 line long c code, there is a ~200 line length simplifyed c code being produced. i assume that from this, around 1000 instruction long binary code will be compiled. this means 32 kbyte code for executing a 60 line long generic c code. however this can be bigger or slower.
when comparing to x86, where 10.000 line C code creates around 150 kbyte of binary
x86: generally 15 byte binary code per line (with gcc)
my subleq compiler: 533 byte binary code per line
note that most of it used as some kind of local stack in my case, so we can say that x86 (+ gcc) is generally 20x more effective than my subleq architecture.
since i am expecting a 5-10x slower general execution compared to x86 aniway, this does not make it more slower. but i will need a code optimizer later, that decrases the memory demand. i alreday figured out, how to do one, but the first goal is to make it working, and optimize only after that. i hope the memory demand is decrasable with 50%.
but my original suggestion to use 4 kbyte l1 cache seems bad idea, i must lift this number up to 8 kbyte temporally. i will measure the number of the cache misses later.