onednn-src 0.1.13

Source of oneAPI Deep Neural Network Library (oneDNN)
Documentation 
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
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95

/*******************************************************************************
* Copyright 2021 Intel Corporation
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
*     http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*******************************************************************************/

#ifndef CPU_X64_MATMUL_BRGEMM_MATMUL_COPY_UTILS_HPP
#define CPU_X64_MATMUL_BRGEMM_MATMUL_COPY_UTILS_HPP

#include "cpu/x64/matmul/brgemm_matmul_utils.hpp"

namespace dnnl {
namespace impl {
namespace cpu {
namespace x64 {
namespace matmul {

struct jit_brgemm_matmul_copy_b_t {
    struct ctx_t {
        const void *src = nullptr;
        const void *tr_src = nullptr;
        const void *compensation_ptr = nullptr;
        const void *zp_a_compensation_ptr = nullptr;
        const void *zp_a_neg_value_ptr = nullptr;
        const void *zp_b_value_ptr = nullptr;
        const void *src_scales_ptr = nullptr;
        const void *wei_scales_ptr = nullptr;

        dim_t current_K_start = 0;
        dim_t current_K_iters = 0;
        dim_t current_K_pad = 0;
        dim_t current_N_blk = 0;
        dim_t dynamic_src_stride = 0;
    };

    virtual void operator()(ctx_t *ctx) = 0;
    virtual status_t create_kernel() = 0;

    jit_brgemm_matmul_copy_b_t(const brgemm_matmul_conf_t *conf)
        : conf_(conf) {}
    virtual ~jit_brgemm_matmul_copy_b_t() = default;

    const brgemm_matmul_conf_t *conf_;
};

struct jit_brgemm_matmul_copy_a_t {
    struct ctx_t {
        const void *src = nullptr;
        const void *tr_src = nullptr;
        const void *zp_b_compensation_buffer_ptr = nullptr;
        const void *zp_a_compensation_result_ptr = nullptr;
        const void *zp_b_neg_val_ptr = nullptr;
        const void *zp_ab_comp_ptr = nullptr;

        dim_t current_K_start = 0;
        dim_t current_K_blk = 0;
        dim_t current_M_blk = 0;
        dim_t dynamic_src_ld = 0;
    };

    virtual void operator()(ctx_t *ctx) = 0;
    virtual status_t create_kernel() = 0;

    jit_brgemm_matmul_copy_a_t(const brgemm_matmul_conf_t *conf)
        : conf_(conf) {}
    virtual ~jit_brgemm_matmul_copy_a_t() = default;

    const brgemm_matmul_conf_t *conf_;
};

status_t create_brgemm_matmul_copy_b(
        std::unique_ptr<jit_brgemm_matmul_copy_b_t> &copy_ker,
        const brgemm_matmul_conf_t *conf);

status_t create_brgemm_matmul_copy_a(
        std::unique_ptr<jit_brgemm_matmul_copy_a_t> &copy_ker,
        const brgemm_matmul_conf_t *conf);

} // namespace matmul
} // namespace x64
} // namespace cpu
} // namespace impl
} // namespace dnnl

#endif