onednn-src 0.1.13

Source of oneAPI Deep Neural Network Library (oneDNN)
Documentation
/*******************************************************************************
* Copyright 2019 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_IP_REF_HPP
#define GPU_INTEL_IP_REF_HPP

#include "common/c_types_map.hpp"
#include "gpu/intel/ip/config.hpp"
#include "gpu/intel/primitive.hpp"

namespace dnnl {
namespace impl {
namespace gpu {
namespace intel {
namespace ip {

struct ref_fwd_t : public primitive_t {
    using primitive_t::primitive_t;
    struct pd_t : public fwd_pd_t {
        using fwd_pd_t::fwd_pd_t;

        DECLARE_COMMON_PD_T("ocl:ref:any", ref_fwd_t);

        status_t init(impl::engine_t *engine) {
            using namespace data_type;
            using namespace prop_kind;
            using namespace data_type;
            assert(engine->kind() == engine_kind::gpu);
            auto *intel_engine = utils::downcast<intel::engine_t *>(engine);

            const auto attr_skip_mask = primitive_attr_t::skip_mask_t::scales
                    | primitive_attr_t::skip_mask_t::post_ops;

            VDISPATCH_INNER_PRODUCT(
                    utils::one_of(desc()->prop_kind, forward_training,
                            forward_inference),
                    VERBOSE_BAD_PROPKIND);

            VDISPATCH_INNER_PRODUCT_SC(
                    set_default_params(), VERBOSE_UNSUPPORTED_TAG);

            auto src_dt = src_md()->data_type;
            auto dst_dt = dst_md()->data_type;
            auto wei_dt = weights_md(0)->data_type;

            const bool is_f32 = src_dt == f32
                    && utils::one_of(wei_dt, f32, s8, u8)
                    && utils::one_of(dst_dt, f32, f16, bf16);
            const bool is_f16 = src_dt == f16
                    && utils::one_of(wei_dt, f16, s8, u8)
                    && utils::one_of(dst_dt, u8, s8, f16, bf16);
            const bool is_bf16 = src_dt == bf16
                    && utils::one_of(wei_dt, bf16, s8, u8)
                    && utils::one_of(dst_dt, bf16, f32);
            const bool is_int8 = utils::one_of(src_dt, u8, s8)
                    && utils::one_of(wei_dt, u8, s8)
                    && utils::one_of(dst_dt, f32, s8, u8, s32, f16, bf16);
            VDISPATCH_INNER_PRODUCT((is_int8 || is_f32 || is_f16 || is_bf16),
                    VERBOSE_UNSUPPORTED_DT);

            VDISPATCH_INNER_PRODUCT(
                    IMPLICATION(with_bias(),
                            utils::one_of(desc()->bias_desc.data_type, u8, s8,
                                    bf16, f16, f32)),
                    VERBOSE_UNSUPPORTED_DT);
            VDISPATCH_INNER_PRODUCT(attr()->has_default_values(attr_skip_mask),
                    VERBOSE_UNSUPPORTED_ATTR);
            VDISPATCH_INNER_PRODUCT(
                    post_ops_with_binary_ok(attr(), desc()->dst_desc),
                    VERBOSE_UNSUPPORTED_POSTOP);
            VDISPATCH_INNER_PRODUCT_SC(attr_.set_default_formats(dst_md(0)),
                    VERBOSE_UNSUPPORTED_POSTOP);
            VDISPATCH_INNER_PRODUCT(
                    IMPLICATION(!attr()->scales_.has_default_values(),
                            utils::one_of(src_md_.data_type, s8, u8)
                                    && attr_scales_ok()),
                    VERBOSE_UNSUPPORTED_SCALES_CFG);
            VDISPATCH_INNER_PRODUCT(
                    IMPLICATION(desc()->src_desc.data_type == f16,
                            intel_engine->mayiuse(
                                    compute::device_ext_t::khr_fp16)),
                    VERBOSE_UNSUPPORTED_DT_CFG);
            CHECK(init_conf(engine));
            return status::success;
        }

        status_t init_conf(impl::engine_t *engine);
        status_t init_kernel_ctx(compute::kernel_ctx_t &kernel_ctx) const;

        conf_t conf;
        offsets_t off;
    };

    status_t init(impl::engine_t *engine) override {
        compute::kernel_ctx_t kernel_ctx;
        status_t status = pd()->init_kernel_ctx(kernel_ctx);
        CHECK(status);

        CHECK(create_kernel(
                engine, &kernel_, "ref_inner_product_fwd", kernel_ctx));
        if (!kernel_) return status::runtime_error;

        return status::success;
    }

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

private:
    status_t execute_forward(const exec_ctx_t &ctx) const;
    const pd_t *pd() const { return (const pd_t *)primitive_t::pd().get(); }
    compute::kernel_t kernel_;
};

struct ref_bwd_data_t : public primitive_t {
    using primitive_t::primitive_t;
    struct pd_t : public bwd_data_pd_t {
        using bwd_data_pd_t::bwd_data_pd_t;

        DECLARE_COMMON_PD_T("ref:any", ref_bwd_data_t);

        status_t init(impl::engine_t *engine) {
            using namespace data_type;
            using namespace prop_kind;
            assert(engine->kind() == engine_kind::gpu);

            VDISPATCH_INNER_PRODUCT(utils::one_of(this->desc()->prop_kind,
                                            backward, backward_data),
                    VERBOSE_BAD_PROPKIND);
            VDISPATCH_INNER_PRODUCT_SC(
                    this->set_default_params(), VERBOSE_UNSUPPORTED_TAG);
            VDISPATCH_INNER_PRODUCT(
                    utils::one_of(true,
                            expect_data_types(
                                    f16, f16, data_type::undef, f16, f32),
                            expect_data_types(
                                    f32, f16, data_type::undef, f16, f32),
                            expect_data_types(
                                    bf16, bf16, data_type::undef, bf16, f32),
                            expect_data_types(
                                    f32, bf16, data_type::undef, bf16, f32),
                            expect_data_types(
                                    f32, f32, data_type::undef, f32, f32)),
                    VERBOSE_UNSUPPORTED_DT);
            VDISPATCH_INNER_PRODUCT(
                    attr()->has_default_values(), VERBOSE_UNSUPPORTED_ATTR);
            CHECK(init_conf(engine));
            return status::success;
        }

        status_t init_conf(impl::engine_t *engine);
        status_t init_kernel_ctx(compute::kernel_ctx_t &kernel_ctx) const;

        conf_t conf;
        offsets_t off;
    };

    status_t init(impl::engine_t *engine) override {
        compute::kernel_ctx_t kernel_ctx;
        status_t status = pd()->init_kernel_ctx(kernel_ctx);
        CHECK(status);

        CHECK(create_kernel(
                engine, &kernel_, "ref_inner_product_bwd_data", kernel_ctx));
        if (!kernel_) return status::runtime_error;

        return status::success;
    }

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

private:
    status_t execute_backward_data(const exec_ctx_t &ctx) const;
    const pd_t *pd() const { return (const pd_t *)primitive_t::pd().get(); }
    compute::kernel_t kernel_;
};

struct ref_bwd_weights_t : public primitive_t {
    using primitive_t::primitive_t;
    struct pd_t : public bwd_weights_pd_t {
        using bwd_weights_pd_t::bwd_weights_pd_t;

        DECLARE_COMMON_PD_T("ref:any", ref_bwd_weights_t);

        status_t init(impl::engine_t *engine) {
            using namespace data_type;
            using namespace prop_kind;
            assert(engine->kind() == engine_kind::gpu);

            VDISPATCH_INNER_PRODUCT(utils::one_of(this->desc()->prop_kind,
                                            backward, backward_weights),
                    VERBOSE_BAD_PROPKIND);
            VDISPATCH_INNER_PRODUCT_SC(
                    this->set_default_params(), VERBOSE_UNSUPPORTED_TAG);
            VDISPATCH_INNER_PRODUCT(
                    utils::one_of(true,
                            expect_data_types(
                                    f16, f16, data_type::undef, f16, f32),
                            expect_data_types(
                                    f16, f32, data_type::undef, f16, f32),
                            expect_data_types(
                                    bf16, bf16, data_type::undef, bf16, f32),
                            expect_data_types(
                                    bf16, f32, data_type::undef, bf16, f32),
                            expect_data_types(
                                    f32, f32, data_type::undef, f32, f32)),
                    VERBOSE_UNSUPPORTED_DT);
            VDISPATCH_INNER_PRODUCT(
                    attr()->has_default_values(), VERBOSE_UNSUPPORTED_ATTR);
            CHECK(init_conf(engine));
            return status::success;
        }

        status_t init_conf(impl::engine_t *engine);
        status_t init_kernel_ctx(compute::kernel_ctx_t &kernel_ctx) const;

        conf_t conf;
        offsets_t off;
    };

    status_t init(impl::engine_t *engine) override {
        compute::kernel_ctx_t kernel_ctx;
        status_t status = pd()->init_kernel_ctx(kernel_ctx);
        CHECK(status);

        CHECK(create_kernel(
                engine, &kernel_, "ref_inner_product_bwd_weights", kernel_ctx));
        if (!kernel_) return status::runtime_error;

        return status::success;
    }

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

private:
    status_t execute_backward_weights(const exec_ctx_t &ctx) const;
    const pd_t *pd() const { return (const pd_t *)primitive_t::pd().get(); }
    compute::kernel_t kernel_;
};

} // namespace ip
} // namespace intel
} // namespace gpu
} // namespace impl
} // namespace dnnl

#endif

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