I've been wondering about the possible performance/missed-optimization
impact of my patch for PR middle-end/98420 and similar IEEE correctness
fixes that disable constant folding optimizations when worrying about -0.0.
In the common situation where the floating point result is used by a
FP comparison, there's no distinction between +0.0 and -0.0, so some
HONOR_SIGNED_ZEROS optimizations that we'd usually disable, are safe.
Consider the following interesting example:
int foo(int x, double y) {
return (x * 0.0) < y;
}
Although we know that x (when converted to double) can't be NaN or Inf,
we still worry that for negative values of x that (x * 0.0) may be -0.0
and so perform the multiplication at run-time. But in this case, the
result of the comparison (-0.0 < y) will be exactly the same as (+0.0 < y)
for any y, hence the above may be safely constant folded to "0.0 < y"
avoiding the multiplication at run-time.
This patch has been tested on x86_64-pc-linux-gnu with make bootstrap
and make -k check with no new failures, and allows GCC to continue to
optimize cases that we optimized in GCC 11 (without regard to correctness).
Ok for mainline?
2022-03-14 Roger Sayle <[email protected]>
gcc/ChangeLog
* match.pd (X CMP (Y-Y) -> X CMP 0.0): New transformation.
(X CMP (Y * 0.0) -> X CMP 0.0): Likewise.
(X CMP X -> true): Test tree_expr_maybe_nan_p instead of HONOR_NANS.
(X LTGT X -> false): Enable if X is not tree_expr_maybe_nan_p, as
this can't trap/signal.
gcc/testsuite/ChangeLog
* gcc.dg/fold-compare-9.c: New test case.
Thanks in advance,
Roger
--
diff --git a/gcc/match.pd b/gcc/match.pd
index 8b44b5c..f3f7865 100644
--- a/gcc/match.pd
+++ b/gcc/match.pd
@@ -4671,6 +4671,35 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
(if (single_use (@2))
(cmp @0 @1)))))
+/* Floating point comparisons can ignore signed zeros. */
+(for cmp (tcc_comparison)
+ /* Simplify (X - X) CMP Y even with -frounding-math. */
+ (simplify
+ (cmp (minus @0 @0) @1)
+ (if (FLOAT_TYPE_P (TREE_TYPE (@0))
+ && !tree_expr_maybe_nan_p (@0)
+ && !tree_expr_maybe_infinite_p (@0))
+ (cmp { build_zero_cst (TREE_TYPE (@0)); } @1)))
+ /* Simplify X CMP (Y - Y) even with -frounding-math. */
+ (simplify
+ (cmp @0 (minus @1 @1))
+ (if (FLOAT_TYPE_P (TREE_TYPE (@1))
+ && !tree_expr_maybe_nan_p (@1)
+ && !tree_expr_maybe_infinite_p (@1))
+ (cmp @0 { build_zero_cst (TREE_TYPE (@1)); })))
+ /* Simplify (X * 0.0) CMP Y. */
+ (simplify
+ (cmp (mult @0 real_zerop@1) @2)
+ (if (!tree_expr_maybe_nan_p (@0)
+ && !tree_expr_maybe_infinite_p (@0))
+ (cmp @1 @2)))
+ /* Simplify X CMP (Y * 0.0). */
+ (simplify
+ (cmp @0 (mult @1 real_zerop@2))
+ (if (!tree_expr_maybe_nan_p (@1)
+ && !tree_expr_maybe_infinite_p (@0))
+ (cmp @0 @2))))
+
/* Simplify (x < 0) ^ (y < 0) to (x ^ y) < 0 and
(x >= 0) ^ (y >= 0) to (x ^ y) < 0. */
(for cmp (lt ge)
@@ -4743,7 +4772,7 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
(simplify
(cmp @0 @0)
(if (! FLOAT_TYPE_P (TREE_TYPE (@0))
- || ! HONOR_NANS (@0))
+ || ! tree_expr_maybe_nan_p (@0))
{ constant_boolean_node (true, type); }
(if (cmp != EQ_EXPR
/* With -ftrapping-math conversion to EQ loses an exception. */
@@ -4755,7 +4784,7 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
(cmp @0 @0)
(if (cmp != NE_EXPR
|| ! FLOAT_TYPE_P (TREE_TYPE (@0))
- || ! HONOR_NANS (@0))
+ || ! tree_expr_maybe_nan_p (@0))
{ constant_boolean_node (false, type); })))
(for cmp (unle unge uneq)
(simplify
@@ -4767,7 +4796,7 @@ DEFINE_INT_AND_FLOAT_ROUND_FN (RINT)
(unordered @0 @0)))
(simplify
(ltgt @0 @0)
- (if (!flag_trapping_math)
+ (if (!flag_trapping_math || !tree_expr_maybe_nan_p (@0))
{ constant_boolean_node (false, type); }))
/* x == ~x -> false */
diff --git a/gcc/testsuite/gcc.dg/fold-compare-9.c
b/gcc/testsuite/gcc.dg/fold-compare-9.c
new file mode 100644
index 0000000..e56f63d
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/fold-compare-9.c
@@ -0,0 +1,8 @@
+/* { dg-do compile } */
+/* { dg-options "-O2 -fdump-tree-optimized" } */
+
+int foo(int x, double y) {
+ return (x * 0.0) < y;
+}
+
+/* { dg-final { scan-tree-dump " y_\[0-9\]\\(D\\) > 0.0" "optimized" } } */
