reg+reg addressing mode on SPARC

[Joe Buck]

Dan Nicolaescu writes:
> Here is the result that I get for the code from the 2340 bug report
> using a version of gcc patched to have ADDRESS_COST 2 for the reg+reg
> addressing modes and 1 for all the other cases:
> 
> ./a.out
> test      absolute   additions      ratio with
> number    time       per second     test0
> 
>  0        0.31sec    161.29M         1.00
>  1        0.31sec    161.29M         1.00
[ all others are one as well ]

Cool.

Hmm.  Seems the next step is to try your patch on other benchmarks, such
as Spec-[year of your choice].  If it looks good, then, well, since I
don't know that much about the backend, I hope that those who do will
seriously look at making this change.