On Sun, 2 Sep 2018, Bernhard Reutner-Fischer wrote:
On 30 August 2018 22:20:06 CEST, Marc Glisse wrote:
Hello,
INTEGRALS_SIGN_PREC_MATCH: the name doesn't really match the semantics.
+#define INTEGRALS_SIGN_PREC_MATCH(A, B) \
+ TYPE_PRECISION (TREE_TYPE (A)) == TYPE_PRECISION (TREE_TYPE
On 30 August 2018 22:20:06 CEST, Marc Glisse wrote:
>Hello,
>
>INTEGRALS_SIGN_PREC_MATCH: the name doesn't really match the semantics.
+#define INTEGRALS_SIGN_PREC_MATCH(A, B) \
+ TYPE_PRECISION (TREE_TYPE (A)) == TYPE_PRECISION (TREE_TYPE (B)) \
+ (TYPE_PRECISION (TREE_TYPE (A)) > TYPE_PRECIS
Hello,
why not put these right next to the one that does x==0 & y==0?
INTEGRALS_SIGN_PREC_MATCH: the name doesn't really match the semantics.
You may need to add a bunch of :s.
I always wonder if such transformations should be blocked until a late
gimple pass so we keep the more natural form
Sorry, I attached a patch file with an minor bug.
Kai
Hi,
this patch implements some folding patterns about integral comparison of
condition. I noticed the lack of folding in gcc by comparing gcc with llvm
result. These optimization are interesting as they are helping to linearize
some condition
i,
this patch implements some folding patterns about integral comparison of
condition. I noticed the lack of folding in gcc by comparing gcc with llvm
result. These optimization are interesting as they are helping to linearize
some conditional expressions. The following patterns are implemented:
