(vec_cond @0 (view_convert? (uncond_op@4 @1 @2)) @3)
(with { tree op_type = TREE_TYPE (@4); }
(if (vectorized_internal_fn_supported_p (as_internal_fn (cond_op), op_type)
- && is_truth_type_for (op_type, TREE_TYPE (@0)))
+ && is_truth_type_for (op_type, TREE_TYPE (@0))
+ && single_use (@4))
(view_convert (cond_op @0 @1 @2 (view_convert:op_type @3))))))
(simplify
(vec_cond @0 @1 (view_convert? (uncond_op@4 @2 @3)))
(with { tree op_type = TREE_TYPE (@4); }
(if (vectorized_internal_fn_supported_p (as_internal_fn (cond_op), op_type)
- && is_truth_type_for (op_type, TREE_TYPE (@0)))
+ && is_truth_type_for (op_type, TREE_TYPE (@0))
+ && single_use (@4))
(view_convert (cond_op (bit_not @0) @2 @3 (view_convert:op_type @1)))))))
/* Same for ternary operations. */
(vec_cond @0 (view_convert? (uncond_op@5 @1 @2 @3)) @4)
(with { tree op_type = TREE_TYPE (@5); }
(if (vectorized_internal_fn_supported_p (as_internal_fn (cond_op), op_type)
- && is_truth_type_for (op_type, TREE_TYPE (@0)))
+ && is_truth_type_for (op_type, TREE_TYPE (@0))
+ && single_use (@5))
(view_convert (cond_op @0 @1 @2 @3 (view_convert:op_type @4))))))
(simplify
(vec_cond @0 @1 (view_convert? (uncond_op@5 @2 @3 @4)))
(with { tree op_type = TREE_TYPE (@5); }
(if (vectorized_internal_fn_supported_p (as_internal_fn (cond_op), op_type)
- && is_truth_type_for (op_type, TREE_TYPE (@0)))
+ && is_truth_type_for (op_type, TREE_TYPE (@0))
+ && single_use (@5))
(view_convert (cond_op (bit_not @0) @2 @3 @4
(view_convert:op_type @1)))))))
#endif
/* { dg-do compile } */
/* { dg-options "-O2 -march=skylake-avx512 -fdump-tree-optimized -DTYPE=int16" } */
-/* { dg-final { scan-tree-dump-times ".COND_SHR" 1 "optimized" } } */
-/* { dg-final { scan-tree-dump-times ".COND_SHL" 1 "optimized" } } */
+/* { dg-final { scan-tree-dump-times "\.COND_" 4 "optimized" } } */
/* { dg-final { scan-assembler-times "vpsraw" 1 } } */
/* { dg-final { scan-assembler-times "vpsllw" 1 } } */
--- /dev/null
+/* { dg-do compile } */
+/* { dg-options "-march=icelake-server -Ofast -fdump-tree-optimized" } */
+/* { dg-final { scan-tree-dump-not "\.COND_SHR" "optimized" } } */
+/* { dg-final { scan-tree-dump-not "\.COND_FMA" "optimized" } } */
+
+void
+cond_shr (unsigned int* __restrict dst,
+ unsigned int* __restrict src,
+ unsigned int* __restrict y,
+ int i_width)
+{
+ for(int x = 0; x < i_width; x++)
+ {
+ unsigned int temp = src[x] >> 3;
+ dst[x] = temp > 255 ? temp : y[x];
+ }
+}
+
+
+void
+cond_fma (float* __restrict dst,
+ float* __restrict src1,
+ float* __restrict src2,
+ float* __restrict src3,
+ unsigned int* __restrict y,
+ int i_width)
+{
+ for(int x = 0; x < i_width; x++)
+ {
+ float temp = __builtin_fmaf (src1[x], src2[x], src3[x]);
+ dst[x] = temp > 0.0f ? temp : y[x];
+ }
+}