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
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136

/*******************************************************************************
* Copyright 2024 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 GPU_INTEL_CONCAT_SIMPLE_HPP
#define GPU_INTEL_CONCAT_SIMPLE_HPP

#include "gpu/intel/concat/config.hpp"
#include "gpu/intel/primitive.hpp"

namespace dnnl {
namespace impl {
namespace gpu {
namespace intel {
namespace concat {

struct simple_params_t : trivially_serializable_t<simple_params_t> {

    const std::vector<const char *> &get_kernel_names() const {
        static const std::vector<const char *> kernel_names
                = {"reusable_simple_concat", "internal_padding_block_concat2"};
        return kernel_names;
    }

    status_t create_generator(const intel::engine_t &engine,
            compute::kernel_bundle_t &bundle) const {
        auto status = engine.create_kernel_bundle(
                bundle, get_kernel_names(), get_kernel_ctx());
        return status;
    }

    compute::kernel_ctx_t get_kernel_ctx() const;

    dim_t n_blocks;
    dim_t blocks[6] = {0};
    dim_t strides[6] = {0};

    dim_t read_block;
    dim_t write_block;
    int n;
    int simd;
    int data_type_size;
    int bytes_per_workitem = 0;
    bool use_large_index = true;
    bool use_internal_padding_kernel = false;
    bool require_stateless_addressing = true;
    uint8_t padding[5] = {0};
};

struct simple_runtime_params_t {
    dim_t dst_extern_dim_size;
    dim_t dst_offset0;
    dim_t src_extern_dim_sizes[64];
    dim_t offset[64];
    dim_t padded_offset[64];

    dim_t dst_concat_axis;
    dim_t dst_padded_concat_axis;
    dim_t read_overlap;
    dim_t gws0_block;
    dim_t inner_axis;

    dim_t src_concat_axis0;
    dim_t src_concat_axis1;
    dim_t padded_src_concat_axis0;
    dim_t padded_src_concat_axis1;

    compute::range_t gws_d = compute::range_t::one();
    compute::range_t lws_d;
};

struct simple_t : public primitive_t {
    using primitive_t::primitive_t;
    struct pd_t : public concat::pd_t {
        using concat::pd_t::pd_t;

        DECLARE_CONCAT_PD_T("ocl:simple:reusable", simple_t);

        status_t init(impl::engine_t *engine) {
            VDISPATCH_CONCAT(n_inputs() <= 64, VERBOSE_BAD_PARAM, "n_inputs");
            VDISPATCH_CONCAT(
                    attr()->has_default_values(), VERBOSE_UNSUPPORTED_ATTR);
            VDISPATCH_CONCAT_SC(set_default_params(), VERBOSE_UNSUPPORTED_TAG);

            CHECK(init_conf(engine));

            return status::success;
        }

        status_t init_conf(impl::engine_t *engine);

        simple_params_t conf;
        simple_runtime_params_t rt_conf;
    };

    status_t init(impl::engine_t *engine) override {
        std::vector<compute::kernel_t> kernels;
        CHECK(create_kernels(
                engine, kernels, pd()->conf.get_kernel_names(), pd()->conf));
        kernel_ = kernels[0];
        internal_padding_kernel_ = kernels[1];

        return status::success;
    }

    status_t execute(const exec_ctx_t &ctx) const override {
        return execute_concat(ctx);
    }

private:
    status_t execute_concat(const exec_ctx_t &ctx) const;
    const pd_t *pd() const { return (const pd_t *)primitive_t::pd().get(); }

    compute::kernel_t kernel_;
    compute::kernel_t internal_padding_kernel_;
};

} // namespace concat
} // namespace intel
} // namespace gpu
} // namespace impl
} // namespace dnnl

#endif