can you further optimize assembly code by changing the binary directly?
are there any known cases where this was necessary?
Just cut the tire open
>>58314908
1. Hardly
2. Probably
Compilers are really good at writing and opyimizing the ASM for you, so there's very little cases one would benefit from it.
Some trivial (like hello world) or edge case (see 16K demoscene) stuff can be made a fraction smaller/faster by removing the overhead that a compiler would introduce.
Example:
Someone rewrote busybox in ASM, its called asmutils and its reportedly a bit faster and smaller.
>>58314908
it's cut where you can't see
>>58314918
the point is how did it get there in the first place
That road looks like the kind where garbage is always laying everywhere and it never gets cleaned up. I wouldn't be surprised if that tire sat there for as long as the tree has been growing.
>>58317869
Someone cut it. It's cut behind the tree...
:)
Either someone cut it from behind or the tree grew with the tire
OP seems like you're thinking that using a lower level of abstraction will result in better performance.
This is not going to be the case going from asm to machine code (binary). asm has a one to one mapping with your cpu's instruction set. Writing straight machine code will end up with the same result as asm, only it will be much harder to write. It may save you from having to compile but is that really worth it?