Re: [PATCH] Variable Expansion in unroller + Register Pressure Utility

[Daniel Berlin <dberlin at dberlin dot org>]

On Sep 22, 2004, at 12:05 PM, Revital Eres wrote:

> > Now I see that Revital added function
> > estimate_register_pressure with a register as parameter which I don't
> > understand.  It just returns difference between number of available 
> > hard
>
> > registers in class (which the pseudo-register belongs to for register
> > pressure estimation purpose) and maximal register pressure in some
> > blocks.  Even if we are trying to create a pseudo-register in the 
> > basic
> > blocks, the returned value says nothing because the new 
> > pseudo-register
> > could be used in some point where the register pressure does not reach
> > maximal value.  So with my point of view it is not accurate evaluation
> > to decide create or not the new pseudo-register.
> >
> > Vlad
>
> This is why among others it is highly conservative approach.
> I can change it to use per-block register pressure check
> if you think it is more accurate.

I think you meant to say that you are willing to change it if it 
actually produces better code that way :)

That's of course, the easiest way to measure what we should be using.
The results of the most accurate analysis in the world are useless if 
it doesn't actually help produce better code.
--Dan