| /external/gemmlowp/public/ |
| D | gemmlowp.h | 40 const OutputPipelineType& output_pipeline) { in GemmWithOutputPipelinePC() argument
42 context, lhs, rhs, result, lhs_offset, rhs_offset, output_pipeline); in GemmWithOutputPipelinePC()
58 const OutputPipelineType& output_pipeline) { in GemmWithOutputPipeline() argument
65 output_pipeline); in GemmWithOutputPipeline()
|
| /external/tensorflow/tensorflow/lite/kernels/ |
| D | cpu_backend_gemm_gemmlowp.h | 112 auto output_pipeline = std::make_tuple(
117 output_pipeline);
119 auto output_pipeline =
124 output_pipeline);
174 auto output_pipeline = std::make_tuple(bias_addition_stage, scale_stage,
179 -lhs_params.zero_point, -rhs_params.zero_point, output_pipeline);
|
| /external/gemmlowp/internal/ |
| D | unpack.h | 137 const OutputPipelineType& output_pipeline) { in UnpackResult() argument
161 output_pipeline_executor_1x1(output_pipeline); in UnpackResult()
163 output_pipeline_executor_4x1(output_pipeline); in UnpackResult()
165 output_pipeline_executor_8x1(output_pipeline); in UnpackResult()
167 output_pipeline_executor_1x4(output_pipeline); in UnpackResult()
169 output_pipeline_executor_4x4(output_pipeline); in UnpackResult()
171 output_pipeline_executor_8x4(output_pipeline); in UnpackResult()
|
| D | multi_thread_gemm.h | 487 output_pipeline(_output_pipeline) {} in GemmWithPackedRhsTask()
519 rhs_offset.block(curr_result_block.start_col, cs), output_pipeline); in Run()
535 const OutputPipelineType& output_pipeline; member
641 const OutputPipelineType& output_pipeline) { in MultiThreadGemm() argument
664 output_pipeline); in MultiThreadGemm()
710 lhs_offset, rhs_offset, block_params, output_pipeline)); in MultiThreadGemm()
|
| D | single_thread_gemm.h | 72 const OutputPipelineType& output_pipeline) { in SingleThreadGemm() argument
148 lhs_offset.block(r, rs), rhs_offset.block(c, cs), output_pipeline); in SingleThreadGemm()
|
| D | dispatch_gemm_shape.h | 178 const OutputPipelineType& output_pipeline) {
196 TransposeTuple(output_pipeline));
202 lhs_offset, rhs_offset, output_pipeline);
|
| D | output.h | 480 OutputPipelineEvalImpl(const OutputPipelineType& output_pipeline)
481 : head_impl(std::get<FirstStage>(output_pipeline)),
482 tail_impl(output_pipeline) {}
538 OutputPipelineExecutor(const OutputPipelineType& output_pipeline)
539 : output_pipeline_eval_impl_(output_pipeline) {}
|
| /external/gemmlowp/doc/ |
| D | public.md | 40 const OutputPipelineType& output_pipeline);
49 &uint8_result_matrix, lhs_offset, rhs_offset, output_pipeline);
75 See below explanation of the `output_pipeline` parameter, and
99 operations on the result, see `output_pipeline`.
100 * `output_pipeline` is a `std::tuple` of output stages (see
|
| D | quantization_example.cc | 356 const auto& output_pipeline = in main() local
364 &actual_uint8_result_map, lhs_offset, rhs_offset, output_pipeline); in main()
|
| /external/gemmlowp/test/ |
| D | benchmark_all_sizes.cc | 138 const auto output_pipeline = in benchmark_8bit() local
144 -128, output_pipeline); in benchmark_8bit()
155 -128, output_pipeline); in benchmark_8bit()
|
| D | test.cc | 780 auto output_pipeline = MakeStandardOutputPipeline<VectorShape::Col>( in TestWithSmallDataPerChannelQuantization() local
785 output_pipeline); in TestWithSmallDataPerChannelQuantization()
977 auto output_pipeline = MakeStandardOutputPipeline<VectorShape::Col>( in TestWithLargeDataPerChannelQuantization() local
982 output_pipeline); in TestWithLargeDataPerChannelQuantization()
1090 auto output_pipeline = MakeStandardOutputPipeline<VectorShape::Col>( in TestMultithreadedPerChannelQuantization() local
1095 output_pipeline); in TestMultithreadedPerChannelQuantization()
|
| /external/ruy/ruy/ |
| D | test.h | 837 const auto& output_pipeline =
843 -lhs.zero_point(), -rhs.zero_point(), output_pipeline);
847 const auto& output_pipeline =
853 -lhs.zero_point(), -rhs.zero_point(), output_pipeline);
858 const auto& output_pipeline = std::make_tuple(
863 -lhs.zero_point(), -rhs.zero_point(), output_pipeline);
867 const auto& output_pipeline = std::make_tuple(
872 -lhs.zero_point(), -rhs.zero_point(), output_pipeline);
|
| /external/tensorflow/tensorflow/lite/kernels/internal/optimized/ |
| D | legacy_optimized_ops.h | 1140 const auto& output_pipeline = GemmlowpOutputPipeline::MakeExp( in FullyConnected() local
1146 filter_offset, input_offset, output_pipeline); in FullyConnected()
1712 auto output_pipeline = in FullyConnected() local
1718 filter_offset, input_offset, output_pipeline); in FullyConnected()
2157 const auto& output_pipeline = GemmlowpOutputPipelineInt8::MakeExp( in FullyConnected() local
2164 filter_offset, input_offset, output_pipeline); in FullyConnected()
2720 const auto& output_pipeline = GemmlowpOutputPipeline::MakeExp( in Conv() local
2726 filter_offset, input_offset, output_pipeline); in Conv()
2898 const auto& output_pipeline = GemmlowpOutputPipeline::MakeExp( in ConvAsGemm() local
2904 filter_offset, input_offset, output_pipeline); in ConvAsGemm()
[all …]
|