1/****************************************************************************
2 * Copyright (C) 2017 Intel Corporation.   All Rights Reserved.
3 *
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
10 *
11 * The above copyright notice and this permission notice (including the next
12 * paragraph) shall be included in all copies or substantial portions of the
13 * Software.
14 *
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
18 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
20 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
21 * IN THE SOFTWARE.
22 ****************************************************************************/
23#if !defined(__SIMD_LIB_AVX2_HPP__)
24#error Do not include this file directly, use "simdlib.hpp" instead.
25#endif
26
27//============================================================================
28// SIMD4 AVX (2) implementation
29//
30// Since this implementation inherits from the AVX (1) implementation,
31// the only operations below ones that replace AVX (1) operations.
32// Only 2 shifts and 2 gathers were introduced with AVX 2
33// Also, add native support for FMA operations
34//============================================================================
35#define SIMD_WRAPPER_3(op) \
36    static SIMDINLINE Float SIMDCALL op(Float a, Float b, Float c) { return _mm_##op(a, b, c); }
37
38SIMD_WRAPPER_3(fmadd_ps); // return (a * b) + c
39SIMD_WRAPPER_3(fmsub_ps); // return (a * b) - c
40
41static SIMDINLINE Integer SIMDCALL sllv_epi32(Integer vA, Integer vB) // return a << b      (uint32)
42{
43    return _mm_sllv_epi32(vA, vB);
44}
45
46static SIMDINLINE Integer SIMDCALL srlv_epi32(Integer vA, Integer vB) // return a >> b      (uint32)
47{
48    return _mm_srlv_epi32(vA, vB);
49}
50
51template <ScaleFactor ScaleT = ScaleFactor::SF_1>
52static SIMDINLINE Float SIMDCALL
53                        i32gather_ps(float const* p, Integer idx) // return *(float*)(((int8*)p) + (idx * ScaleT))
54{
55    return _mm_i32gather_ps(p, idx, static_cast<const int>(ScaleT));
56}
57
58// for each element: (mask & (1 << 31)) ? (i32gather_ps<ScaleT>(p, idx), mask = 0) : old
59template <ScaleFactor ScaleT = ScaleFactor::SF_1>
60static SIMDINLINE Float SIMDCALL
61                        mask_i32gather_ps(Float old, float const* p, Integer idx, Float mask)
62{
63    return _mm_mask_i32gather_ps(old, p, idx, mask, static_cast<const int>(ScaleT));
64}
65
66#undef SIMD_WRAPPER_3
67