Hi all,
I am playing around comparing the performance of a Java/C/C# implementation of 
an algorithm that is rate-limiting in an analysis pipeline.  I was hoping to 
track down the differences between them a bit more clearly by looking at the 
assembly.
I gather there are at least two options to view assembly.
1 - Call mono with -v -v options, which produces a readable file but with no 
comments indicating which part of the method you are looking at.
2- Call mono --aot which produces a .s file.  Unfortunately I can't read it 
though because it seems to be in a byte format and not the typical instruction 
type that I am used to.
I gather option 1 is the preferred way to view assembly, but thought I would 
ask if anyone knew some tips and tricks for making this process easier (or 
alternate ways).
Also, I noticed according to this: 
http://www.itu.dk/~sestoft/papers/numericperformance.pdf
That C# seems to be using the fmultp and faddp rather than the mulsd and addsd 
instructions (which they indicate are faster).  Does anyone know if there is a 
way to compile down to use these instructions for floating point calculations?
Thanks for any help,Nigel                                         
_______________________________________________
Mono-list maillist  -  [email protected]
http://lists.ximian.com/mailman/listinfo/mono-list

Reply via email to