03:56:26
##forth
<
lisbeths>
so crc my VM has come to a head for now. next I'm gonna be writing some compilers that are much more static: https://github.com/memesmith0/vm/blob/main/vm.c if you are gonna read the compiler outut please do it on x86_64 with the recommended commpilation command in the comment. also if you have time pls check it on the tcc compiler with maximum runtime-filesize-saving optimizations. the former is MUCH faster at about 6 k. the
03:56:26
##forth
<lisbeths>
latter is around 4k but can probably dip lower, especially of hand compiled to asm. I am hoping you can automate a message getting sent to me when I eventually log in again, or if you want pls email me your response to this design at memsmith.kestrel@gmail.com the key thing is just to be purely register based like a processor with a minimal design and its focus is to get into reader macros as quickly and simply as possible to
08:17:10
##forth
<lisbeths>
so the ai has convinced me that for microcontrollers the way the truth and the light is essentially heavily heavily hand optimized rtos, which at that point im honestly just gonna hand compile to asm. its not hard onve ive got it prototyped. and then pretty much on Unix machines, which I know you guys don't agree with me on how to code them for high level stuff: i basically need to take a totally different path for maximum
08:17:10
##forth
<lisbeths>
probability. so my VM is gonna split between mcu code and unix code. the mcu one obviously is gonna be asm. probably gonna simulate a rtos dev board slowly cause i don't have cash. and then for the Unix one the main thing is I am switching from malloc and free straight to mmap of char array and munmap and I am just automatically sharing all of the memory with other processes in /tmp for lazy concurrency. and then I might add an
23:05:59
##forth
<ajhidd>
forthBot: S" 3I/Atlas "S IMAGE
23:06:35
##forth
* ajhidd giggles