On Thu, Nov 25, 2010 at 3:32 AM, Dave Korn <dave.korn.cyg...@gmail.com> wrote: > On 24/11/2010 14:17, Richard Guenther wrote: > >> I don't see why RTL optimizers should be different from tree optimizers. > > I thought half the point of tree-ssa in the first place was to separate > optimisation out from target-specific stuff and do it on an independent level? > > On 24/11/2010 15:32, Richard Guenther wrote: >> As we are moving towards doing more target dependent optimizations >> on the tree level this doesn't sound like a sustainable opinion. > > Wait, we're doing that? Isn't that the same mistake we made earlier?
GIMPLE SSA is a much easier IL to work on compared to RTL and we are (slowly) moving (early) RTL optimizations towards the tree level (see recent changes around fma, widening multiplication, etc.). And we do have target dependent passes on trees, like the vectorizer, induction variable optimization, various foldings that depend on target availability of builtins, etc. Even all trees for decls and types are target dependent. > On 24/11/2010 14:17, Richard Guenther wrote: >> And we don't want to pay the overhead of hookization every target >> dependent constant just for the odd guys who want multi-target >> compilers that have those constants differing. > > Why not? Precisely how big is this cost? Back in the old days we all used > to want to avoid virtual functions, because of the cost of a > function-call-through-pointer, but that certainly isn't justified any more and > may not even have been then. Well. I don't see value in adding even small cost if there is no benefit. I'm trying to guess what people think the benefit would be, as I think they are just confused as of what a completely hookized backend would allow you to do (nothing!). >> a multi-target compiler where the hooks are in shared loadable >> modules > > It's not just Diego who envisions that, I think it would be an excellent > long-term goal too. And I thought that was why all the work to hookize macros > was motivated in the first place. What's the benefit to our users of Diegos vision? The downside will be a slower compiler (maybe not by much, but we get that 1-3% slowdown every release and it adds up and people are complaining). Richard. > cheers, > DaveK > >
