[PATCH] D24805: [GVNSink] Initial GVNSink prototype

[Daniel Berlin via llvm-commits]

On Wed, Sep 21, 2016 at 3:43 PM, Matthias Braun <matze at braunis.de> wrote:

> MatzeB added a subscriber: MatzeB.
> MatzeB added a comment.
>
> Without looking at anything in the patch. For the example given, wouldn't
> it be enough to create some DAGCombine style rules:


> x = PHI(OP(y, c), OP(z, c))
>  =>
>  x = op(PHI(z, z), c)
>
> ?


There is a set of rewrite rules you can prove cover some of the
equivalences, but you can also prove it will not  cover all of them.
See https://people.eecs.berkeley.edu/~necula/Papers/gvndet_sas04.pdf 's
discussion of the RKS algorithm.

(which includes a discussion of this rewrite rule)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160921/99ae1da4/attachment.html>