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
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192

/*******************************************************************************
* Copyright 2016 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.
*******************************************************************************/

#include "common/c_types_map.hpp"
#include "common/dnnl_thread.hpp"
#include "common/type_helpers.hpp"

#include "cpu/binary_injector_utils.hpp"
#include "cpu/cpu_primitive.hpp"
#include "cpu/gemm_inner_product.hpp"

namespace dnnl {
namespace impl {
namespace cpu {

using namespace dnnl::impl::status;
using namespace dnnl::impl::prop_kind;
using namespace dnnl::impl::data_type;
using namespace dnnl::impl::format_tag;
using namespace dnnl::impl::primitive_kind;

template <impl::data_type_t data_type>
status_t gemm_inner_product_fwd_t<data_type>::execute_forward(
        const exec_ctx_t &ctx) const {
    auto src = CTX_IN_MEM(const data_t *, DNNL_ARG_SRC);
    auto weights = CTX_IN_MEM(const data_t *, DNNL_ARG_WEIGHTS);
    auto bias = CTX_IN_MEM(const data_t *, DNNL_ARG_BIAS);
    auto dst = CTX_OUT_MEM(data_t *, DNNL_ARG_DST);
    const auto post_ops_binary_rhs_arg_vec
            = binary_injector_utils::prepare_binary_args(
                    this->pd()->attr()->post_ops_, ctx);

    const auto &scratchpad = ctx.get_scratchpad_grantor();
    auto acc_ptr = scratchpad.template get<data_t>(
            memory_tracking::names::key_gemm_tmp_buffer);
    auto acc = pd()->sum_through_pp_kernel_ ? acc_ptr : dst;

    const dim_t MB = pd()->MB();
    const dim_t OC = pd()->OC();
    const dim_t IC = pd()->IC_total_padded();

    const auto &wmd = *pd()->weights_md();
    const auto &smd = *pd()->src_md();
    // check if OC is NOT the leading dimension
    bool wei_tr = wmd.format_desc.blocking.strides[0] != 1;
    // check if MB is the leading dimension
    bool src_tr = smd.format_desc.blocking.strides[0] == 1 && IC > 1;
    const auto sum_idx = pd()->attr()->post_ops_.find(primitive_kind::sum);
    float beta = !pd()->sum_through_pp_kernel_ && sum_idx >= 0
            ? pd()->attr()->post_ops_.entry_[sum_idx].sum.scale
            : 0.f;

    float alpha = 1.;
    status_t st = extended_sgemm(wei_tr ? "T" : "N", src_tr ? "T" : "N", &OC,
            &MB, &IC, &alpha, weights, wei_tr ? &IC : &OC, src,
            src_tr ? &MB : &IC, &beta, acc, &OC,
            postops_in_ip_ ? nullptr : bias);

    if (st != status::success) return st;

    if (postops_in_ip_) {
        const bool force_sequential = pp_kernel_->sequential_kernel();
        parallel(force_sequential ? 1 : 0, [&](int ithr, int nthr) {
            size_t start, end;
            balance211((size_t)(OC * MB), nthr, ithr, start, end);
            const size_t dim1_off = start % OC;
            (*pp_kernel_)(dst, acc, (char *)bias, nullptr, 1.0f, start, start,
                    dim1_off, end, 0,
                    pd()->OC() * pd()->OD() * pd()->OH() * pd()->OW(), nullptr,
                    post_ops_binary_rhs_arg_vec.data(), dst, 0, ctx,
                    *pd()->dst_md());
        });
    }

    return status::success;
}

template <impl::data_type_t data_type>
status_t gemm_inner_product_bwd_data_t<data_type>::execute_backward_data(
        const exec_ctx_t &ctx) const {
    auto diff_dst = CTX_IN_MEM(const data_t *, DNNL_ARG_DIFF_DST);
    auto weights = CTX_IN_MEM(const data_t *, DNNL_ARG_WEIGHTS);
    auto diff_src = CTX_OUT_MEM(data_t *, DNNL_ARG_DIFF_SRC);

    const dim_t MB = pd()->MB();
    const dim_t OC = pd()->OC();
    const dim_t IC = pd()->IC_total_padded();

    const auto &wmd = *pd()->weights_md();
    const auto &smd = *pd()->diff_src_md();
    bool wei_tr = wmd.format_desc.blocking.strides[0] == 1;
    // check if MB is the leading dimension
    bool dsrc_tr = smd.format_desc.blocking.strides[0] == 1 && IC > 1;

    float alpha = 1.0, beta = 0.0;
    status_t st = status::success;
    if (dsrc_tr)
        st = extended_sgemm(wei_tr ? "T" : "N", "N", &OC, &IC, &MB, &alpha,
                diff_dst, &OC, weights, wei_tr ? &OC : &IC, &beta, diff_src,
                &MB);
    else
        st = extended_sgemm(wei_tr ? "T" : "N", "N", &IC, &MB, &OC, &alpha,
                weights, wei_tr ? &OC : &IC, diff_dst, &OC, &beta, diff_src,
                &IC);

    return st;
}

template <impl::data_type_t data_type>
status_t gemm_inner_product_bwd_weights_t<data_type>::execute_backward_weights(
        const exec_ctx_t &ctx) const {
    auto diff_dst = CTX_IN_MEM(const data_t *, DNNL_ARG_DIFF_DST);
    auto src = CTX_IN_MEM(const data_t *, DNNL_ARG_SRC);
    auto diff_weights = CTX_OUT_MEM(data_t *, DNNL_ARG_DIFF_WEIGHTS);
    auto diff_bias = CTX_OUT_MEM(data_t *, DNNL_ARG_DIFF_BIAS);

    const memory_desc_wrapper diff_dst_d(pd()->diff_dst_md());
    const memory_desc_wrapper diff_bias_d(pd()->diff_weights_md(1));

    diff_dst += diff_dst_d.offset0();

    const dim_t MB = pd()->MB();
    const dim_t OC = pd()->OC();
    const dim_t IC = pd()->IC_total_padded();

    const auto &wmd = *pd()->diff_weights_md();
    const auto &smd = *pd()->src_md();
    bool wei_tr = wmd.format_desc.blocking.strides[0] == 1;
    // check if MB is the leading dimension
    bool src_tr = smd.format_desc.blocking.strides[0] == 1 && IC > 1;

    float alpha = 1.0, beta = 0.0;
    status_t st = status::success;
    if (wei_tr)
        st = extended_sgemm("N", src_tr ? "N" : "T", &OC, &IC, &MB, &alpha,
                diff_dst, &OC, src, src_tr ? &MB : &IC, &beta, diff_weights,
                &OC);
    else
        st = extended_sgemm("N", src_tr ? "N" : "T", &IC, &OC, &MB, &alpha, src,
                src_tr ? &MB : &IC, diff_dst, &OC, &beta, diff_weights, &IC);

    if (st != status::success) return st;

    if (diff_bias) {
        diff_bias += diff_bias_d.offset0();
        constexpr dim_t blksize = 8;
        const dim_t OC_blocks = utils::div_up(OC, blksize);
        parallel(0, [&](const int ithr, const int nthr) {
            dim_t oc_s {0}, oc_e {0};
            balance211(OC_blocks, nthr, ithr, oc_s, oc_e);
            oc_s = std::min(oc_s * blksize, OC);
            oc_e = std::min(oc_e * blksize, OC);

            PRAGMA_OMP_SIMD()
            for (dim_t oc = oc_s; oc < oc_e; ++oc) {
                diff_bias[oc] = diff_dst[oc];
            }

            for (dim_t mb = 1; mb < MB; ++mb) {
                PRAGMA_OMP_SIMD()
                for (dim_t oc = oc_s; oc < oc_e; ++oc) {
                    diff_bias[oc] += diff_dst[mb * OC + oc];
                }
            }
        });
    }

    return status::success;
}

template struct gemm_inner_product_fwd_t<data_type::f32>;
template struct gemm_inner_product_bwd_data_t<data_type::f32>;
template struct gemm_inner_product_bwd_weights_t<data_type::f32>;

} // namespace cpu
} // namespace impl
} // namespace dnnl

// vim: et ts=4 sw=4 cindent cino+=l0,\:4,N-s