[new-regalloc] What is the status on current sources

[Michael Matz]

Hi,

On Thu, 8 Feb 2001, Geert Bosch wrote:
>   It's been about a week in trying to get them merged properly, work out
>   problems, etc.
>   
>   Michael can successfully bootstrap on x86, and make check shows we are in
>   *very* good shape (over 14000 passes).
>   
>   I can successfully bootstrap on PPC.
>   
>   So just hold on a little longer.
> 
> Thanks for the update, great news! I'll hold my breath... :-)

But at the first time don't expect any performance improvements.  Not in
compile time, which might get horrible, and in situations with lot
spilling (x86 comes to mind) also not in run time, as currently the spill
code is in no way optimized, coalesced moves are not deleted, sometimes
multiple spills/reloads are inserted for the same register, sometimes dead
spills into stack slots just overwritten one insn later happen and so on
;)  I just want to give you a warning ;)  These problems will of course be
fixed, it just takes a little time; at least it basically works.


Ciao,
Michael.