aboutsummaryrefslogtreecommitdiff
path: root/arm_compute/core/NEON/kernels/assembly/kernels/a64_sgemm_12x8.hpp
blob: e229e215efec7be8717fd39dacd1b6cb5ddb0ab4 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
/*
 * Copyright (c) 2017 ARM Limited.
 *
 * SPDX-License-Identifier: MIT
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to
 * deal in the Software without restriction, including without limitation the
 * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
 * sell copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in all
 * copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */
#pragma once

#ifdef __aarch64__

// Actual kernel implementations
#include "a64_sgemm_12x8/generic.hpp"
#include "a64_sgemm_12x8/a53.hpp"

// 12x8 SGEMM "strategy" class.
//
// This describes the characteristics of a family of kernels, in terms of
// the required interleave properties and the output block size.
//
// All kernels in the family must share these characteristics.  The actual
// kernel to be used can be chosen at runtime, based on the CPU_type
// structure.
class sgemm_12x8 {
public:
    typedef float operand_type;
    typedef float result_type;

    typedef void (*kern_type)(const float *, const float *, float *, int, int, int);

    /* Describes the data layout for A input */
    static const int A_interleave = 8;
    static const int A_block = 1;
    static const int A_transpose = 0;

    /* Same for B input */
    static const int B_interleave = 12;
    static const int B_block = 1;
    static const int B_transpose = 1;

    /* Kernel blocking parameters */
    static const int out_width = 12;
    static const int out_height = 8;
    static const int k_unroll = 1;

    kern_type kernel{nullptr};

    sgemm_12x8(const CPUInfo *ci) {
        kernel = a64_sgemm_asimd_12x8;
        if (ci->CPU == CPUTarget::A53) {
            kernel = a64_sgemm_asimd_12x8_a53;
        }
    }
};

#endif // __aarch64__