inference-engine/thirdparty/clDNN/kernel_selector/core/actual_kernels/fully_connected/fully_connected_block_kernel_base.cpp

   1 /*
   2 // Copyright (c) 2016 Intel Corporation
   3 //
   4 // Licensed under the Apache License, Version 2.0 (the "License");
   5 // you may not use this file except in compliance with the License.
   6 // You may obtain a copy of the License at
   7 //
   8 //      http://www.apache.org/licenses/LICENSE-2.0
   9 //
  10 // Unless required by applicable law or agreed to in writing, software
  11 // distributed under the License is distributed on an "AS IS" BASIS,
  12 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13 // See the License for the specific language governing permissions and
  14 // limitations under the License.
  15 */
  16
  17 #include "fully_connected_block_kernel_base.h"
  18
  19 namespace kernel_selector
  20 {
  21     JitConstants FullyConnectedBlockKernelBase::GetJitConstants(const fully_connected_params& params, const FullyConnectedBlockKernelBase::DispatchData& data) const
  22     {
  23         auto cldnnJit = FullyConnectedKernelBase::GetJitConstants(params, data);
  24
  25         const auto batches_per_work_item = GetBatchesPerWorkItem(params);
  26
  27         cldnnJit.AddConstant(MakeJitConstant("NEURONS_PER_WORK_ITEM", GetNeuronsPerWorkItem(params))); // how many neurons for a single batch will a single work item produce
  28         cldnnJit.AddConstant(MakeJitConstant("BATCHES_PER_WORK_ITEM", batches_per_work_item));             // how many batches will a single work item compute
  29         cldnnJit.AddConstant(MakeJitConstant("OUTPUT_ELEMENTS_COUNT", params.output.LogicalSize() / params.output.Batch().v));
  30
  31         return cldnnJit;
  32     }
  33
  34 }