Speedup CSE by 5%

[Daniel Berlin]

On Mon, 2005-01-17 at 16:59 +0100, Arend Bayer wrote:

> > Presumably the real gain here is the inlining of for_each_rtx, not
> > the inlining of approx_reg_cost_1 into approx_reg_cost.  Right?
> 
> If by "inlining for_each_rtx" you include the constant propagation that avoids
> the indirect function call to approx_reg_cost_1, then probably yes.
> 

This sounds like it would be a good testcase for the IPCP people, since
we know it produces a good speedup, and involves IPCP of a function
pointer into a called function.