onednn-src 0.1.13

/*******************************************************************************
* Copyright 2020 Intel Corporation
* Copyright 2020-2022 Codeplay Software Limited
*
* 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/compiler_workarounds.hpp"

#include "gpu/amd/miopen_gemm_inner_product.hpp"
#include "gpu/amd/miopen_inner_product.hpp"
#include "gpu/amd/stream.hpp"
#include "gpu/amd/sycl_hip_scoped_context.hpp"
#include "xpu/sycl/buffer_memory_storage.hpp"
#include "xpu/sycl/memory_storage_helper.hpp"

namespace dnnl {
namespace impl {
namespace gpu {
namespace amd {

status_t miopen_inner_product_fwd_t::execute(const exec_ctx_t &ctx) const {
    if (pd()->has_zero_dim_memory()) return status::success;

    amd::stream_t *hip_stream = utils::downcast<amd::stream_t *>(ctx.stream());

    return hip_stream->interop_task([&](::sycl::handler &cgh) {
        auto arg_src = CTX_IN_SYCL_MEMORY(DNNL_ARG_SRC);
        auto arg_wei = CTX_IN_SYCL_MEMORY(DNNL_ARG_WEIGHTS);
        auto arg_bias = CTX_IN_SYCL_MEMORY(DNNL_ARG_BIAS);
        auto arg_dst = CTX_OUT_SYCL_MEMORY(DNNL_ARG_DST);
        auto arg_ip_scratch = CTX_SCRATCH_SYCL_MEMORY(
                memory_tracking::names::key_iprod_int_dat_in_acc_dt);
        auto arg_spacial_scratch
                = CTX_SCRATCH_SYCL_MEMORY(memory_tracking::names::key_none);
        auto arg_scaled_bias_scratch = CTX_SCRATCH_SYCL_MEMORY(
                memory_tracking::names::key_conv_adjusted_scales);
        compat::host_task(
                cgh, [= COMPAT_THIS_CAPTURE](const compat::interop_handle &ih) {
            auto &sycl_engine
                    = *utils::downcast<amd::engine_t *>(hip_stream->engine());
            auto sc = hip_sycl_scoped_context_handler_t(sycl_engine);
            auto native_stream = hip_stream->get_underlying_stream();
            auto miopen_handle = hip_stream->get_miopen_handle(native_stream);
            auto rocblas_handle = hip_stream->get_rocblas_handle(native_stream);

            std::vector<void *> args;

            args.push_back(arg_src.get_native_pointer(ih));
            args.push_back(arg_wei.get_native_pointer(ih));
            args.push_back(arg_bias.get_native_pointer(ih));
            args.push_back(arg_dst.get_native_pointer(ih));
            args.push_back(arg_ip_scratch.get_native_pointer(ih));
            args.push_back(arg_spacial_scratch.get_native_pointer(ih));
            args.push_back(arg_scaled_bias_scratch.get_native_pointer(ih));

            pd()->inner_product_impl_->execute(
                    miopen_handle, rocblas_handle, args);
        });
    });
}

status_t miopen_inner_product_bwd_data_t::execute(const exec_ctx_t &ctx) const {
    if (pd()->has_zero_dim_memory()) return status::success;
    amd::stream_t *hip_stream = utils::downcast<amd::stream_t *>(ctx.stream());

    return hip_stream->interop_task([&](::sycl::handler &cgh) {
        auto arg_diff_dst = CTX_IN_SYCL_MEMORY(DNNL_ARG_DIFF_DST);
        auto arg_wei = CTX_IN_SYCL_MEMORY(DNNL_ARG_WEIGHTS);
        auto arg_diff_src = CTX_OUT_SYCL_MEMORY(DNNL_ARG_DIFF_SRC);
        auto arg_ip_scratch = CTX_SCRATCH_SYCL_MEMORY(
                memory_tracking::names::key_iprod_int_dat_in_acc_dt);
        auto arg_spacial_scratch
                = CTX_SCRATCH_SYCL_MEMORY(memory_tracking::names::key_none);

        compat::host_task(
                cgh, [= COMPAT_THIS_CAPTURE](const compat::interop_handle &ih) {
            auto &sycl_engine
                    = *utils::downcast<amd::engine_t *>(hip_stream->engine());
            auto sc = hip_sycl_scoped_context_handler_t(sycl_engine);
            auto native_stream = hip_stream->get_underlying_stream();
            auto miopen_handle = hip_stream->get_miopen_handle(native_stream);
            auto rocblas_handle = hip_stream->get_rocblas_handle(native_stream);

            std::vector<void *> args;

            args.push_back(arg_diff_src.get_native_pointer(ih));
            args.push_back(arg_wei.get_native_pointer(ih));
            args.push_back(arg_diff_dst.get_native_pointer(ih));
            args.push_back(arg_ip_scratch.get_native_pointer(ih));
            args.push_back(arg_spacial_scratch.get_native_pointer(ih));

            pd()->inner_product_impl_->execute(
                    miopen_handle, rocblas_handle, args);
        });
    });
}

status_t miopen_inner_product_bwd_weights_t::execute(
        const exec_ctx_t &ctx) const {

    amd::stream_t *hip_stream = utils::downcast<amd::stream_t *>(ctx.stream());

    if (pd()->has_zero_dim_memory()) {
        auto wei_sz = memory_desc_wrapper(pd()->diff_weights_md(0)).size();
        size_t bias_sz = (pd()->with_bias()
                        ? memory_desc_wrapper(pd()->diff_weights_md(1)).size()
                        : 0);

        if (wei_sz != 0) {
            auto status = hip_stream->fill(
                    CTX_OUT_STORAGE(DNNL_ARG_DIFF_WEIGHTS), 0, wei_sz,
                    hip_stream->ctx().get_deps(), hip_stream->ctx().get_deps());
            if (status != status::success) return status;
        }
        if (bias_sz != 0) {
            auto status = hip_stream->fill(CTX_OUT_STORAGE(DNNL_ARG_DIFF_BIAS),
                    0, bias_sz, hip_stream->ctx().get_deps(),
                    hip_stream->ctx().get_deps());
            if (status != status::success) return status;
        }
        return status::success;
    }

    return hip_stream->interop_task([&](::sycl::handler &cgh) {
        auto arg_src = CTX_IN_SYCL_MEMORY(DNNL_ARG_SRC);
        auto arg_diff_dst = CTX_IN_SYCL_MEMORY(DNNL_ARG_DIFF_DST);
        auto arg_diff_wei = CTX_OUT_SYCL_MEMORY(DNNL_ARG_DIFF_WEIGHTS);
        auto arg_bias = CTX_OUT_SYCL_MEMORY(DNNL_ARG_DIFF_BIAS);
        auto arg_ip_scratch = CTX_SCRATCH_SYCL_MEMORY(
                memory_tracking::names::key_iprod_int_dat_in_acc_dt);
        auto arg_spacial_scratch
                = CTX_SCRATCH_SYCL_MEMORY(memory_tracking::names::key_none);

        compat::host_task(
                cgh, [= COMPAT_THIS_CAPTURE](const compat::interop_handle &ih) {
            auto &sycl_engine
                    = *utils::downcast<amd::engine_t *>(hip_stream->engine());
            auto sc = hip_sycl_scoped_context_handler_t(sycl_engine);
            auto native_stream = hip_stream->get_underlying_stream();
            auto miopen_handle = hip_stream->get_miopen_handle(native_stream);
            auto rocblas_handle = hip_stream->get_rocblas_handle(native_stream);
            std::vector<void *> args;

            args.push_back(arg_src.get_native_pointer(ih));
            args.push_back(arg_diff_dst.get_native_pointer(ih));
            args.push_back(arg_diff_wei.get_native_pointer(ih));
            args.push_back(arg_bias.get_native_pointer(ih));
            args.push_back(arg_ip_scratch.get_native_pointer(ih));
            args.push_back(arg_spacial_scratch.get_native_pointer(ih));

            pd()->inner_product_impl_->execute(
                    miopen_handle, rocblas_handle, args);
        });
    });
}

} // namespace amd
} // namespace gpu
} // namespace impl
} // namespace dnnl