1  /* { dg-do compile } */
       2  /* { dg-options "-O2 -ftree-vectorize" } */
       3  
       4  #include <stdint.h>
       5  
       6  #define abs(A) ((A) < 0 ? -(A) : (A))
       7  #define neg(A) (-(A))
       8  
       9  #define DEF_LOOP(TYPE, OP)					\
      10    void __attribute__ ((noipa))					\
      11    test_##TYPE##_##OP (TYPE *__restrict r, TYPE *__restrict a,	\
      12  		      TYPE *__restrict pred, int n)		\
      13    {								\
      14      for (int i = 0; i < n; ++i)					\
      15        r[i] = pred[i] ? OP (a[i]) : 0;				\
      16    }
      17  
      18  #define TEST_INT_TYPE(T, TYPE) \
      19    T (TYPE, abs) \
      20    T (TYPE, neg)
      21  
      22  #define TEST_FLOAT_TYPE(T, TYPE, SUFFIX) \
      23    T (TYPE, __builtin_fabs##SUFFIX) \
      24    T (TYPE, neg)
      25  
      26  #define TEST_ALL(T) \
      27    TEST_INT_TYPE (T, int8_t) \
      28    TEST_INT_TYPE (T, int16_t) \
      29    TEST_INT_TYPE (T, int32_t) \
      30    TEST_INT_TYPE (T, int64_t) \
      31    TEST_FLOAT_TYPE (T, _Float16, f16) \
      32    TEST_FLOAT_TYPE (T, float, f) \
      33    TEST_FLOAT_TYPE (T, double, )
      34  
      35  TEST_ALL (DEF_LOOP)
      36  
      37  /* { dg-final { scan-assembler-times {\tabs\tz[0-9]+\.b, p[0-7]/m,} 1 } } */
      38  /* { dg-final { scan-assembler-times {\tabs\tz[0-9]+\.h, p[0-7]/m,} 1 } } */
      39  /* { dg-final { scan-assembler-times {\tabs\tz[0-9]+\.s, p[0-7]/m,} 1 } } */
      40  /* { dg-final { scan-assembler-times {\tabs\tz[0-9]+\.d, p[0-7]/m,} 1 } } */
      41  
      42  /* { dg-final { scan-assembler-times {\tneg\tz[0-9]+\.b, p[0-7]/m,} 1 } } */
      43  /* { dg-final { scan-assembler-times {\tneg\tz[0-9]+\.h, p[0-7]/m,} 1 } } */
      44  /* { dg-final { scan-assembler-times {\tneg\tz[0-9]+\.s, p[0-7]/m,} 1 } } */
      45  /* { dg-final { scan-assembler-times {\tneg\tz[0-9]+\.d, p[0-7]/m,} 1 } } */
      46  
      47  /* { dg-final { scan-assembler-times {\tfabs\tz[0-9]+\.h, p[0-7]/m,} 1 } } */
      48  /* { dg-final { scan-assembler-times {\tfabs\tz[0-9]+\.s, p[0-7]/m,} 1 } } */
      49  /* { dg-final { scan-assembler-times {\tfabs\tz[0-9]+\.d, p[0-7]/m,} 1 } } */
      50  
      51  /* { dg-final { scan-assembler-times {\tfneg\tz[0-9]+\.h, p[0-7]/m,} 1 } } */
      52  /* { dg-final { scan-assembler-times {\tfneg\tz[0-9]+\.s, p[0-7]/m,} 1 } } */
      53  /* { dg-final { scan-assembler-times {\tfneg\tz[0-9]+\.d, p[0-7]/m,} 1 } } */
      54  
      55  /* Really we should be able to use MOVPRFX /z here, but at the moment
      56     we're relying on combine to merge a SEL and an arithmetic operation,
      57     and the SEL doesn't allow the "false" value to be zero when the "true"
      58     value is a register.  */
      59  /* { dg-final { scan-assembler-times {\tmovprfx\tz[0-9]+, z[0-9]+\n} 7 } } */
      60  /* { dg-final { scan-assembler-times {\tmovprfx\tz[0-9]+\.b, p[0-7]/z, z[0-9]+\.b} 1 } } */
      61  /* { dg-final { scan-assembler-times {\tmovprfx\tz[0-9]+\.h, p[0-7]/z, z[0-9]+\.h} 2 } } */
      62  /* { dg-final { scan-assembler-times {\tmovprfx\tz[0-9]+\.s, p[0-7]/z, z[0-9]+\.s} 2 } } */
      63  /* { dg-final { scan-assembler-times {\tmovprfx\tz[0-9]+\.d, p[0-7]/z, z[0-9]+\.d} 2 } } */
      64  
      65  /* { dg-final { scan-assembler-not {\tmov\tz[^\n]*z} } } */
      66  /* { dg-final { scan-assembler-not {\tsel\t} } } */