2 // Copyright (c) 2016 Intel Corporation
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
8 // http://www.apache.org/licenses/LICENSE-2.0
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.
17 #include "fully_connected_kernel_bfyx_ref.h"
18 #include "kernel_selector_utils.h"
20 namespace kernel_selector
22 ParamsKey FullyConnected_bfyx_Ref::GetSupportedKey() const
25 k.EnableInputDataType(Datatype::F16);
26 k.EnableInputDataType(Datatype::F32);
27 k.EnableInputDataType(Datatype::INT8);
28 k.EnableOutputDataType(Datatype::F16);
29 k.EnableOutputDataType(Datatype::F32);
30 k.EnableOutputDataType(Datatype::INT8);
31 k.EnableInputWeightsType(WeightsType::F16);
32 k.EnableInputWeightsType(WeightsType::F32);
33 k.EnableInputWeightsType(WeightsType::INT8);
34 k.EnableAllInputLayout();
35 k.EnableInputLayout(DataLayout::bf);
36 k.EnableOutputLayout(DataLayout::bf);
37 k.EnableOutputLayout(DataLayout::fb);
38 k.EnableBiasPerOutput();
39 k.EnableBiasPerFeature();
40 k.EnableNonBiasTerm();
41 k.EnableTensorOffset();
42 k.EnableTensorPitches();
44 k.EnableInt8Quantization();
45 k.EnableOutputCalibration();
49 FullyConnected_bfyx_Ref::DispatchData FullyConnected_bfyx_Ref::SetDefault(const fully_connected_params& params, int ) const
51 auto runInfo = Parent::SetDefault(params);
53 std::vector<size_t> global = { params.output.Feature().v, params.output.Batch().v };
54 std::vector<size_t> local = GetOptimalLocalWorkGroupSizes(global);
56 runInfo.gws0 = global[0];
57 runInfo.gws1 = global[1];
60 runInfo.lws0 = local[0];
61 runInfo.lws1 = local[1];
67 KernelsData FullyConnected_bfyx_Ref::GetKernelsData(const Params& params, const optional_params& options) const
70 for (size_t i = 0; i < autoTuneOptions.size(); i++)
72 KernelsData kd = GetTunedKernelsDataByIndex(params, options, DataLayout::bfyx,
73 { WeightsLayout::oiyx, WeightsLayout::oyxi, WeightsLayout::iyxo, WeightsLayout::yxio },
74 DONT_USE_IF_HAVE_SOMETHING_ELSE, (int)i);
77 res.emplace_back(kd[0]);