Home
last modified time | relevance | path

Searched refs:copy_done (Results 1 – 23 of 23) sorted by relevance

/external/tensorflow/tensorflow/lite/kernels/internal/optimized/
Ddepthwiseconv_uint8_3x3_filter.h6834 int copy_done = 0;
6847 copy_done += 15;
6851 for (; (copy_done + 16) <= copy_size; copy_done += 16) {
6853 util_vld1q_x8(input_block_data + input_block_offset + copy_done);
6857 TFLITE_DCHECK_EQ((start_width + copy_done) % 16, 0);
6859 copy_done);
6860 vst1q_s8(scratch_data + start_width + copy_done, work_reg);
6863 if (copy_done + 8 <= copy_size) {
6865 util_vld1_x8(input_block_data + input_block_offset + copy_done);
6869 TFLITE_DCHECK_EQ((start_width + copy_done) % 8, 0);
[all …]
Ddepthwiseconv_uint8_transitional.h1027 int copy_done = 0;
1039 copy_done += 15;
1043 for (; (copy_done + 16) <= copy_size; copy_done += 16) {
1044 memcpy(tmp_data, input_block_data + input_block_offset + copy_done,
1049 TFLITE_DCHECK_EQ((start_width + copy_done) % 16, 0);
1050 memcpy(&scratch_data[start_width + copy_done], tmp_data, 16);
1053 const int copy_remaining = copy_size - copy_done;
1061 if (copy_done < copy_size_adjusted) {
1065 input_block_data + input_block_offset + copy_done -
1080 width_overall_micro_repeats - (start_width + copy_done) / 4;
[all …]
/external/tensorflow/tensorflow/compiler/xla/service/
Dhlo_matchers.cc259 const HloInstruction* copy_done = instruction; in MatchAndExplain() local
260 if (!copy_done->shape().has_layout()) { in MatchAndExplain()
265 if (copy_done->shape().layout().memory_space() != to_space_) { in MatchAndExplain()
267 << copy_done->shape().layout().memory_space() << ", expected " in MatchAndExplain()
273 copy_done->operands()[0]->operands()[0]; in MatchAndExplain()
Dhlo_matchers_test.cc283 auto copy_done = HloInstruction::CreateUnary( in TEST_F() local
286 EXPECT_THAT(copy_done.get(), op::AsyncCopy(2, 1, op::Parameter(0))); in TEST_F()
291 EXPECT_THAT(Explain(copy_done.get(), op::AsyncCopy(3, 1, op::Parameter(0))), in TEST_F()
296 EXPECT_THAT(Explain(copy_done.get(), op::AsyncCopy(2, 3, op::Parameter(0))), in TEST_F()
Dlogical_buffer_analysis.h66 Status HandleCopyDone(HloInstruction* copy_done) override;
Ddfs_hlo_visitor_with_default.h219 Status HandleCopyDone(HloInstructionPtr copy_done) override { in HandleCopyDone() argument
220 return DefaultAction(copy_done); in HandleCopyDone()
Dtuple_points_to_analysis.cc344 Status TuplePointsToAnalysis::HandleCopyDone(HloInstruction* copy_done) { in HandleCopyDone() argument
346 PointsToSet& points_to_set = CreateEmptyPointsToSet(copy_done); in HandleCopyDone()
348 GetPointsToSet(copy_done->operand(0)); in HandleCopyDone()
Dhlo_dataflow_analysis.h224 bool UpdateCopyDoneValueSet(HloInstruction* copy_done);
Dlogical_buffer_analysis.cc173 Status LogicalBufferAnalysis::HandleCopyDone(HloInstruction* copy_done) { in HandleCopyDone() argument
Dtuple_points_to_analysis_test.cc342 auto copy_done = builder.AddInstruction(HloInstruction::CreateUnary( in TEST_F() local
349 EXPECT_FALSE(points_to_analysis_->GetPointsToSet(copy_done).IsAmbiguous()); in TEST_F()
350 EXPECT_TRUE(points_to_analysis_->GetPointsToSet(copy_done).IsDistinct()); in TEST_F()
355 ExpectHasBufferAliases(copy_start, {0}, {{copy_start, {0}}, {copy_done, {}}}); in TEST_F()
Dhlo_dataflow_analysis.cc465 bool HloDataflowAnalysis::UpdateCopyDoneValueSet(HloInstruction* copy_done) { in UpdateCopyDoneValueSet() argument
466 CHECK_EQ(copy_done->opcode(), HloOpcode::kCopyDone); in UpdateCopyDoneValueSet()
470 GetValueSet(copy_done->operand(0), {0}); in UpdateCopyDoneValueSet()
471 HloValueSet& value_set = GetValueSet(copy_done); in UpdateCopyDoneValueSet()
Dtuple_points_to_analysis.h254 Status HandleCopyDone(HloInstruction* copy_done) override;
Dhlo_verifier.cc991 Status ShapeVerifier::HandleCopyDone(HloInstruction* copy_done) { in HandleCopyDone() argument
992 const Shape& operand_shape = copy_done->operand(0)->shape(); in HandleCopyDone()
1002 copy_done->ToString()); in HandleCopyDone()
1004 return CheckShape(copy_done, ShapeUtil::GetTupleElementShape( in HandleCopyDone()
1005 copy_done->operand(0)->shape(), 0)); in HandleCopyDone()
Dhlo_evaluator.h233 Status HandleCopyDone(HloInstruction* copy_done) override;
Dhlo_verifier.h99 Status HandleCopyDone(HloInstruction* copy_done) override;
Ddfs_hlo_visitor.h269 virtual Status HandleCopyDone(HloInstructionPtr copy_done) = 0;
Dhlo_evaluator.cc1899 Status HloEvaluator::HandleCopyDone(HloInstruction* copy_done) { in HandleCopyDone() argument
1900 const HloInstruction* operand = copy_done->operand(0); in HandleCopyDone()
1909 evaluated_[copy_done] = in HandleCopyDone()
1911 TF_RETURN_IF_ERROR(evaluated_[copy_done].CopyFrom(operand_tuple_literal, in HandleCopyDone()
Dmemory_space_assignment.h618 HloInstruction* copy_done() const { return copy_done_; } in copy_done() function
Dmemory_space_assignment.cc721 copy_start() == other.copy_start() && copy_done() == other.copy_done(); in operator ==()
3173 copy_allocation->copy_done()); in ScheduleAsynchronousCopies()
Dhlo_dataflow_analysis_test.cc1236 auto copy_done = builder.AddInstruction(HloInstruction::CreateUnary( in TEST_P() local
1250 EXPECT_FALSE(analysis.ValueIsDefinedAt(copy_done, /*index=*/{})); in TEST_P()
1252 HloValuesAt(copy_done, /*index=*/{}), in TEST_P()
/external/tensorflow/tensorflow/core/kernels/image/
Dnon_max_suppression_op.cu.cc352 gpuEvent_t copy_done; in NmsGpu() local
354 gpuEventCreateWithFlags(&copy_done, gpuEventDisableTiming)); in NmsGpu()
357 TF_RETURN_IF_CUDA_ERROR(gpuEventRecord(copy_done, device.stream())); in NmsGpu()
358 TF_RETURN_IF_CUDA_ERROR(gpuEventSynchronize(copy_done)); in NmsGpu()
360 gpuEventDestroy(copy_done); in NmsGpu()
394 gpuEvent_t copy_done; in CountIf() local
396 gpuEventCreateWithFlags(&copy_done, gpuEventDisableTiming)); in CountIf()
403 TF_RETURN_IF_CUDA_ERROR(gpuEventRecord(copy_done, device.stream())); in CountIf()
404 TF_RETURN_IF_CUDA_ERROR(gpuEventSynchronize(copy_done)); in CountIf()
Dgenerate_box_proposals_op.cu.cc467 gpuEvent_t copy_done; in Compute() local
468 gpuEventCreate(&copy_done); in Compute()
509 gpuEventRecord(copy_done, d.stream())); in Compute()
510 TF_OP_REQUIRES_CUDA_SUCCESS(context, gpuEventSynchronize(copy_done)); in Compute()
/external/tensorflow/tensorflow/core/common_runtime/device/
Ddevice_event_mgr_test.cc323 int64 copy_done = 0; member
335 ts.compute_done = ts.compute_done - ts.copy_done; in DisplayTimes()
336 ts.copy_done = ts.copy_done - ts.start; in DisplayTimes()
351 << " copy: " << times->at(i).copy_done in DisplayTimes()
398 times->at(r).copy_done = Env::Default()->NowMicros(); in DoAddChain()