1  /* { dg-do compile }  */
       2  /* { dg-require-effective-target arm_v8_2a_fp16_neon_ok }  */
       3  /* { dg-options "-O2 -fno-fast-math" }  */
       4  /* { dg-add-options arm_v8_2a_fp16_neon }  */
       5  
       6  /* Test instructions generated for half-precision arithmetic without
       7     unsafe-math-optimizations.  */
       8  
       9  typedef _Float16 float16_t;
      10  typedef _Float16 float16x4_t __attribute__ ((vector_size (8)));
      11  typedef _Float16 float16x8_t __attribute__ ((vector_size (16)));
      12  
      13  typedef short int16x4_t __attribute__ ((vector_size (8)));
      14  typedef short int int16x8_t  __attribute__ ((vector_size (16)));
      15  
      16  float16_t
      17  fp16_abs (float16_t a)
      18  {
      19    return __builtin_fabsf16 (a);
      20  }
      21  
      22  #define TEST_UNOP(NAME, OPERATOR, TY)		\
      23    TY test_##NAME##_##TY (TY a)			\
      24    {						\
      25      return OPERATOR (a);			\
      26    }
      27  
      28  #define TEST_BINOP(NAME, OPERATOR, TY)		\
      29    TY test_##NAME##_##TY (TY a, TY b)		\
      30    {						\
      31      return a OPERATOR b;			\
      32    }
      33  
      34  #define TEST_CMP(NAME, OPERATOR, RTY, TY)	\
      35    RTY test_##NAME##_##TY (TY a, TY b)		\
      36    {						\
      37      return a OPERATOR b;			\
      38    }
      39  
      40  /* Scalars.  */
      41  
      42  TEST_UNOP (neg, -, float16_t)
      43  TEST_UNOP (abs, fp16_abs, float16_t)
      44  
      45  TEST_BINOP (add, +, float16_t)
      46  TEST_BINOP (sub, -, float16_t)
      47  TEST_BINOP (mult, *, float16_t)
      48  TEST_BINOP (div, /, float16_t)
      49  
      50  TEST_CMP (equal, ==, int, float16_t)
      51  TEST_CMP (unequal, !=, int, float16_t)
      52  TEST_CMP (lessthan, <, int, float16_t)
      53  TEST_CMP (greaterthan, >, int, float16_t)
      54  TEST_CMP (lessthanequal, <=, int, float16_t)
      55  TEST_CMP (greaterthanqual, >=, int, float16_t)
      56  
      57  /* Vectors of size 4.  */
      58  
      59  TEST_UNOP (neg, -, float16x4_t)
      60  
      61  TEST_BINOP (add, +, float16x4_t)
      62  TEST_BINOP (sub, -, float16x4_t)
      63  TEST_BINOP (mult, *, float16x4_t)
      64  TEST_BINOP (div, /, float16x4_t)
      65  
      66  TEST_CMP (equal, ==, int16x4_t, float16x4_t)
      67  TEST_CMP (unequal, !=, int16x4_t, float16x4_t)
      68  TEST_CMP (lessthan, <, int16x4_t, float16x4_t)
      69  TEST_CMP (greaterthan, >, int16x4_t, float16x4_t)
      70  TEST_CMP (lessthanequal, <=, int16x4_t, float16x4_t)
      71  TEST_CMP (greaterthanqual, >=, int16x4_t, float16x4_t)
      72  
      73  /* Vectors of size 8.  */
      74  
      75  TEST_UNOP (neg, -, float16x8_t)
      76  
      77  TEST_BINOP (add, +, float16x8_t)
      78  TEST_BINOP (sub, -, float16x8_t)
      79  TEST_BINOP (mult, *, float16x8_t)
      80  TEST_BINOP (div, /, float16x8_t)
      81  
      82  TEST_CMP (equal, ==, int16x8_t, float16x8_t)
      83  TEST_CMP (unequal, !=, int16x8_t, float16x8_t)
      84  TEST_CMP (lessthan, <, int16x8_t, float16x8_t)
      85  TEST_CMP (greaterthan, >, int16x8_t, float16x8_t)
      86  TEST_CMP (lessthanequal, <=, int16x8_t, float16x8_t)
      87  TEST_CMP (greaterthanqual, >=, int16x8_t, float16x8_t)
      88  
      89  /* { dg-final { scan-assembler-times {vneg\.f16\ts[0-9]+, s[0-9]+} 1 } }  */
      90  /* { dg-final { scan-assembler-times {vneg\.f16\td[0-9]+, d[0-9]+} 1 } }  */
      91  /* { dg-final { scan-assembler-times {vneg\.f16\tq[0-9]+, q[0-9]+} 1 } }  */
      92  /* { dg-final { scan-assembler-times {vabs\.f16\ts[0-9]+, s[0-9]+} 2 } }  */
      93  
      94  /* { dg-final { scan-assembler-times {vadd\.f16\ts[0-9]+, s[0-9]+, s[0-9]+} 1 } }  */
      95  /* { dg-final { scan-assembler-times {vadd\.f16\td[0-9]+, d[0-9]+, d[0-9]+} 1 } }  */
      96  /* { dg-final { scan-assembler-times {vadd\.f16\tq[0-9]+, q[0-9]+, q[0-9]+} 1 } }  */
      97  
      98  /* { dg-final { scan-assembler-times {vsub\.f16\ts[0-9]+, s[0-9]+, s[0-9]+} 1 } }  */
      99  /* { dg-final { scan-assembler-times {vsub\.f16\td[0-9]+, d[0-9]+, d[0-9]+} 1 } }  */
     100  /* { dg-final { scan-assembler-times {vsub\.f16\tq[0-9]+, q[0-9]+, q[0-9]+} 1 } }  */
     101  
     102  /* { dg-final { scan-assembler-times {vmul\.f16\ts[0-9]+, s[0-9]+, s[0-9]+} 1 } }  */
     103  /* { dg-final { scan-assembler-times {vmul\.f16\td[0-9]+, d[0-9]+, d[0-9]+} 1 } }  */
     104  /* { dg-final { scan-assembler-times {vmul\.f16\tq[0-9]+, q[0-9]+, q[0-9]+} 1 } }  */
     105  
     106  /* { dg-final { scan-assembler-times {vdiv\.f16\ts[0-9]+, s[0-9]+, s[0-9]+} 13 } }  */
     107  /* { dg-final { scan-assembler-times {vcmp\.f32\ts[0-9]+, s[0-9]+} 26 } }  */
     108  /* { dg-final { scan-assembler-times {vcmpe\.f32\ts[0-9]+, s[0-9]+} 52 } }  */
     109  
     110  /* { dg-final { scan-assembler-not {vadd\.f32} } }  */
     111  /* { dg-final { scan-assembler-not {vsub\.f32} } }  */
     112  /* { dg-final { scan-assembler-not {vmul\.f32} } }  */
     113  /* { dg-final { scan-assembler-not {vdiv\.f32} } }  */
     114  /* { dg-final { scan-assembler-not {vcmp\.f16} } }  */
     115  /* { dg-final { scan-assembler-not {vcmpe\.f16} } }  */