This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [tree-ssa] where to fix this?

On Jan 6, 2004, at 7:29 PM, wrote:

In message <>, Daniel Berlin wr
When we follow def-use chains in the dominator optimizer to reassociate
and simplify operands. The stuff we do right now is pretty simple, but
it's one of the two areas of the dominator optimizer that I do expect
will need to be extended.

I tried a variant on this back right before we had renaming SSA. It was a forward and backward substitution pass, which tried to see if any of the resulting expressions were GIMPLE (after trying folding, of course). They never were.
Apparently your code was buggy or you weren't looking at the right stuff.
That's a bit harsh of a statement (not that i'm surprised, however).

However, I was mainly interested in substitution of array indexing expressions, for the purposes of simplifying array access expressions.

I've definitely got cases where were substitution & folding results in
a gimple expression.

Wake me when you've got percentage speed increase or decreased memory usage (due to less expressions) numbers.

Simply the number of expressions eliminated means nothing really by itself.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]