The simplification (outertype)((innertype0)a+(innertype1)b) to
((newtype)a+(newtype)b) ends up using TYPE_PRECISION to check
whether it can elide a conversion but in some paths there can
be VECTOR_TYPEs where this instead compares the number of lanes.
The following fixes the missed optimizations and uses
element_precision in those places.
Bootstrap and regtest ongoing on x86_64-unknown-linux-gnu, will
push after that finished.
* match.pd ((outertype)((innertype0)a+(innertype1)b)
-> ((newtype)a+(newtype)b)): Use element_precision
where appropriate.
---
gcc/match.pd | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/gcc/match.pd b/gcc/match.pd
index 85d562a531d..48b76e6a051 100644
--- a/gcc/match.pd
+++ b/gcc/match.pd
@@ -7428,9 +7428,9 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
&& newtype == type
&& types_match (newtype, type))
(op (convert:newtype @1) (convert:newtype @2))
- (with { if (TYPE_PRECISION (ty1) > TYPE_PRECISION (newtype))
+ (with { if (element_precision (ty1) > element_precision (newtype))
newtype = ty1;
- if (TYPE_PRECISION (ty2) > TYPE_PRECISION (newtype))
+ if (element_precision (ty2) > element_precision (newtype))
newtype = ty2; }
/* Sometimes this transformation is safe (cannot
change results through affecting double rounding
@@ -7453,9 +7453,9 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
exponent range for the product or ratio of two
values representable in the TYPE to be within the
range of normal values of ITYPE. */
- (if (TYPE_PRECISION (newtype) < TYPE_PRECISION (itype)
+ (if (element_precision (newtype) < element_precision (itype)
&& (flag_unsafe_math_optimizations
- || (TYPE_PRECISION (newtype) == TYPE_PRECISION (type)
+ || (element_precision (newtype) == element_precision
(type)
&& real_can_shorten_arithmetic (TYPE_MODE (itype),
TYPE_MODE (type))
&& !excess_precision_type (newtype)))
--
2.35.3
