Slow compilation

[toncho11]

I am using version 0.2 compiling on an Amstrad 1640 with 8Mhz 8086 and 640 kb memory.
ELKS latest 0.9 dev

It seems that it works! But it seems that compilation of the 3 demos takes forever. I think the main culprit is nasm. Just compiling cprintf.asm takes a lot of time. Compiling chess.asm by has still not finished after 25 minutest. After 45 minutes it has not finished the compilation of the 3 demos.

Some links related to NASM:

• https://www.reddit.com/r/Compilers/comments/xx3qke/is_nasm_a_good_assembly_language_to_build_a/
• https://board.flatassembler.net/topic.php?p=73971

@ghaerr

[ghaerr]

Thanks @toncho11 for the testing on your Amstrad!

As I suggested on ghaerr/elks#2144 (comment), I think we're going to have to ditch NASM for ELKS. After completing the arena allocator, I was able to see that NASM performs several thousand calls to malloc to assemble the output of a one-line C program. Just how to it manages to do that is beyond me, but I would say it would not be worth spending time trying to fix such a monster, especially since we're using an ancient version of it anyways. (The newer versions stand no chance of running on ELKS, they're huge). It's another case of a large address space and very fast CPUs hiding very bad design. Oh well.

I'll take a look at your attached links and also look for some other assemblers. In the meantime, we have AS86 working which @rafael2k has reported is very speedy. I was neck deep in the allocator code but now have time to get back to enhancing the tools themselves. I plan on getting C86 working with AS86 in the next few weeks so we can continue moving forward regardless of when we replace NASM. (AS86 and NASM assembly language format differ by quite a bit - we would prefer to find a NASM-source-style-compatible assembler, but don't have to).

[toncho11]

The test hello world did compile indeed and it works. I couldn't wait, it was just staying there after the nasm chess line. The chess.asm is 1933 lines.