1 | /* Copyright (C) 2013-2017 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_SHUFFLE_ZBYTES16_H |
9 | #define LIBSIMDPP_SIMDPP_CORE_SHUFFLE_ZBYTES16_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/get_expr.h> |
17 | #include <simdpp/detail/insn/shuffle_zbytes16.h> |
18 | |
19 | namespace simdpp { |
20 | namespace SIMDPP_ARCH_NAMESPACE { |
21 | |
22 | /** Selects bytes from two vectors according to a mask, optionally selecting |
23 | zero. Each byte within the mask defines which element to select: |
24 | * Bit 7 results in the result byte being zeroed, if set. |
25 | * Bits 6-5 must be zero or the behavior is undefined |
26 | * Bit 4 defines which vector to select. 0 corresponds to @a a, 1 to @a b. |
27 | * Bits 3-0 define the element within the selected vector. |
28 | */ |
29 | template<unsigned N, class V1, class V2, class E3> SIMDPP_INL |
30 | typename detail::get_expr_nomask<V1>::empty |
31 | shuffle_zbytes16(const any_vec8<N,V1>& a, const any_vec8<N,V2>& b, |
32 | const uint8<N,E3>& mask) |
33 | { |
34 | typename detail::get_expr_nomask<V1>::type ra = a.wrapped().eval(), |
35 | rb = b.wrapped().eval(); |
36 | return detail::insn::i_shuffle_zbytes16(ra, rb, mask.eval()); |
37 | } |
38 | |
39 | template<unsigned N, class V1, class V2, class E3> SIMDPP_INL |
40 | typename detail::get_expr_nomask<V1>::empty |
41 | shuffle_zbytes16(const any_vec16<N,V1>& a, const any_vec16<N,V2>& b, |
42 | const uint16<N,E3>& mask) |
43 | { |
44 | typename detail::get_expr_nomask<V1>::type ra = a.wrapped().eval(), |
45 | rb = b.wrapped().eval(); |
46 | return detail::insn::i_shuffle_zbytes16(ra, rb, mask.eval()); |
47 | } |
48 | |
49 | template<unsigned N, class V1, class V2, class E3> SIMDPP_INL |
50 | typename detail::get_expr_nomask<V1>::empty |
51 | shuffle_zbytes16(const any_vec32<N,V1>& a, const any_vec32<N,V2>& b, |
52 | const uint32<N,E3>& mask) |
53 | { |
54 | typename detail::get_expr_nomask<V1>::type ra = a.wrapped().eval(), |
55 | rb = b.wrapped().eval(); |
56 | return detail::insn::i_shuffle_zbytes16(ra, rb, mask.eval()); |
57 | } |
58 | |
59 | template<unsigned N, class V1, class V2, class E3> SIMDPP_INL |
60 | typename detail::get_expr_nomask<V1>::empty |
61 | shuffle_zbytes16(const any_vec64<N,V1>& a, const any_vec64<N,V2>& b, |
62 | const uint64<N,E3>& mask) |
63 | { |
64 | typename detail::get_expr_nomask<V1>::type ra = a.wrapped().eval(), |
65 | rb = b.wrapped().eval(); |
66 | return detail::insn::i_shuffle_zbytes16(ra, rb, mask.eval()); |
67 | } |
68 | |
69 | } // namespace SIMDPP_ARCH_NAMESPACE |
70 | } // namespace simdpp |
71 | |
72 | #endif |
73 | |
74 | |