|assembly optimizer tool firstname.lastname@example.org (Jatin Bhateja) (2005-10-14)|
|RE: assembly optimizer tool email@example.com (Naveen Sharma, Noida) (2005-10-19)|
|From:||"Naveen Sharma, Noida" <firstname.lastname@example.org>|
|Date:||19 Oct 2005 02:38:15 -0400|
|Posted-Date:||19 Oct 2005 02:38:15 EDT|
The effectiveness of traditional optimizations on assembly code is
non-satisfactory w.r.t effort that will go into doing something useful.
I am not sure if you gain anything significant.
Good Instruction scheduling needs good alias analysis (thats a real
problem working with low level intermediate code as assembly). At
assembly level, hard registers will restrict instruction movement
severely as they introduce dependencies. Another issue you have to take
care is span dependent instructions. Plus all ideo-syncracies that a
hardware might have.
> Actually I was thinking that as there are lots of compilers which
> does not produces the optimized assembly code. If at all there
> exists some tool
I think you need to figure out the cause why compiler couldn't emit
good code before assuming a optimizing assembler or a post-object
optimizer will be able to do the job.
Anyways, See this reference for useful tips on what can be done at
[Debray98] S. Debray, R. Muth, and M.Weippert, Alias Analysis of
Executable Code, In The 25th ACM SIGPLAN-SIGACT Symposium on Principles
of ProgrammingLanguages, Orlando, Florida (1998)
Return to the
Search the comp.compilers archives again.