| 1 | /* Copyright (C) 2013-2014 Povilas Kanapickas <povilas@radix.lt> |
| 2 | |
| 3 | Distributed under the Boost Software License, Version 1.0. |
| 4 | (See accompanying file LICENSE_1_0.txt or copy at |
| 5 | http://www.boost.org/LICENSE_1_0.txt) |
| 6 | */ |
| 7 | |
| 8 | #ifndef LIBSIMDPP_SIMDPP_CORE_CMP_GE_H |
| 9 | #define LIBSIMDPP_SIMDPP_CORE_CMP_GE_H |
| 10 | |
| 11 | #ifndef LIBSIMDPP_SIMD_H |
| 12 | #error "This file must be included through simd.h" |
| 13 | #endif |
| 14 | |
| 15 | #include <simdpp/types.h> |
| 16 | #include <simdpp/detail/insn/cmp_ge.h> |
| 17 | #include <simdpp/core/detail/scalar_arg_impl.h> |
| 18 | |
| 19 | namespace simdpp { |
| 20 | namespace SIMDPP_ARCH_NAMESPACE { |
| 21 | |
| 22 | /** Compares the values of two signed int16x8 vectors for greater-than |
| 23 | |
| 24 | @code |
| 25 | r0 = (a0 >= b0) ? ~0x0 : 0x0 |
| 26 | ... |
| 27 | rN = (aN >= bN) ? ~0x0 : 0x0 |
| 28 | @endcode |
| 29 | */ |
| 30 | template<unsigned N, class E1, class E2> SIMDPP_INL |
| 31 | mask_int8<N,expr_empty> cmp_ge(const int8<N,E1>& a, |
| 32 | const int8<N,E2>& b) |
| 33 | { |
| 34 | return detail::insn::i_cmp_ge(a.eval(), b.eval()); |
| 35 | } |
| 36 | |
| 37 | SIMDPP_SCALAR_ARG_IMPL_VEC(cmp_ge, mask_int8, int8) |
| 38 | |
| 39 | template<unsigned N, class E1, class E2> SIMDPP_INL |
| 40 | mask_int8<N,expr_empty> cmp_ge(const uint8<N,E1>& a, |
| 41 | const uint8<N,E2>& b) |
| 42 | { |
| 43 | return detail::insn::i_cmp_ge(a.eval(), b.eval()); |
| 44 | } |
| 45 | |
| 46 | SIMDPP_SCALAR_ARG_IMPL_VEC(cmp_ge, mask_int8, uint8) |
| 47 | |
| 48 | template<unsigned N, class E1, class E2> SIMDPP_INL |
| 49 | mask_int16<N,expr_empty> cmp_ge(const int16<N,E1>& a, |
| 50 | const int16<N,E2>& b) |
| 51 | { |
| 52 | return detail::insn::i_cmp_ge(a.eval(), b.eval()); |
| 53 | } |
| 54 | |
| 55 | SIMDPP_SCALAR_ARG_IMPL_VEC(cmp_ge, mask_int16, int16) |
| 56 | |
| 57 | template<unsigned N, class E1, class E2> SIMDPP_INL |
| 58 | mask_int16<N,expr_empty> cmp_ge(const uint16<N,E1>& a, |
| 59 | const uint16<N,E2>& b) |
| 60 | { |
| 61 | return detail::insn::i_cmp_ge(a.eval(), b.eval()); |
| 62 | } |
| 63 | |
| 64 | SIMDPP_SCALAR_ARG_IMPL_VEC(cmp_ge, mask_int16, uint16) |
| 65 | |
| 66 | template<unsigned N, class E1, class E2> SIMDPP_INL |
| 67 | mask_int32<N,expr_empty> cmp_ge(const int32<N,E1>& a, |
| 68 | const int32<N,E2>& b) |
| 69 | { |
| 70 | return detail::insn::i_cmp_ge(a.eval(), b.eval()); |
| 71 | } |
| 72 | |
| 73 | SIMDPP_SCALAR_ARG_IMPL_VEC(cmp_ge, mask_int32, int32) |
| 74 | |
| 75 | template<unsigned N, class E1, class E2> SIMDPP_INL |
| 76 | mask_int32<N,expr_empty> cmp_ge(const uint32<N,E1>& a, |
| 77 | const uint32<N,E2>& b) |
| 78 | { |
| 79 | return detail::insn::i_cmp_ge(a.eval(), b.eval()); |
| 80 | } |
| 81 | |
| 82 | SIMDPP_SCALAR_ARG_IMPL_VEC(cmp_ge, mask_int32, uint32) |
| 83 | |
| 84 | template<unsigned N, class E1, class E2> SIMDPP_INL |
| 85 | mask_int64<N,expr_empty> cmp_ge(const int64<N,E1>& a, |
| 86 | const int64<N,E2>& b) |
| 87 | { |
| 88 | return detail::insn::i_cmp_ge(a.eval(), b.eval()); |
| 89 | } |
| 90 | |
| 91 | SIMDPP_SCALAR_ARG_IMPL_VEC(cmp_ge, mask_int64, int64) |
| 92 | |
| 93 | template<unsigned N, class E1, class E2> SIMDPP_INL |
| 94 | mask_int64<N,expr_empty> cmp_ge(const uint64<N,E1>& a, |
| 95 | const uint64<N,E2>& b) |
| 96 | { |
| 97 | return detail::insn::i_cmp_ge(a.eval(), b.eval()); |
| 98 | } |
| 99 | SIMDPP_SCALAR_ARG_IMPL_VEC(cmp_ge, mask_int64, uint64) |
| 100 | |
| 101 | /** Compares the values of two float32x4 vectors for greater-than or equal |
| 102 | |
| 103 | @code |
| 104 | r0 = (a0 >= b0) ? 0xffffffff : 0x0 |
| 105 | ... |
| 106 | rN = (aN >= bN) ? 0xffffffff : 0x0 |
| 107 | @endcode |
| 108 | |
| 109 | @par 256-bit version: |
| 110 | @icost{SSE2-SSE4.1, NEON, ALTIVEC, 2} |
| 111 | */ |
| 112 | template<unsigned N, class E1, class E2> SIMDPP_INL |
| 113 | mask_float32<N,expr_empty> cmp_ge(const float32<N,E1>& a, |
| 114 | const float32<N,E2>& b) |
| 115 | { |
| 116 | return detail::insn::i_cmp_ge(a.eval(), b.eval()); |
| 117 | } |
| 118 | |
| 119 | SIMDPP_SCALAR_ARG_IMPL_VEC(cmp_ge, mask_float32, float32) |
| 120 | |
| 121 | /** Compares the values of two float64x2 vectors for greater-than |
| 122 | |
| 123 | @code |
| 124 | r0 = (a0 >= b0) ? 0xffffffffffffffff : 0x0 |
| 125 | ... |
| 126 | rN = (aN >= bN) ? 0xffffffffffffffff : 0x0 |
| 127 | @endcode |
| 128 | |
| 129 | @par 128-bit version: |
| 130 | @novec{NEON, ALTIVEC} |
| 131 | |
| 132 | @par 256-bit version: |
| 133 | @novec{NEON, ALTIVEC} |
| 134 | @icost{SSE2-SSE4.1, 2} |
| 135 | */ |
| 136 | template<unsigned N, class E1, class E2> SIMDPP_INL |
| 137 | mask_float64<N,expr_empty> cmp_ge(const float64<N,E1>& a, |
| 138 | const float64<N,E2>& b) |
| 139 | { |
| 140 | return detail::insn::i_cmp_ge(a.eval(), b.eval()); |
| 141 | } |
| 142 | |
| 143 | SIMDPP_SCALAR_ARG_IMPL_VEC(cmp_ge, mask_float64, float64) |
| 144 | |
| 145 | } // namespace SIMDPP_ARCH_NAMESPACE |
| 146 | } // namespace simdpp |
| 147 | |
| 148 | #endif |
| 149 | |
| 150 | |