| /external/tensorflow/tensorflow/compiler/xla/service/gpu/ |
| D | gpu_fusible_test.cc | 49 ASSERT_EQ(reduce->opcode(), HloOpcode::kReduce); in TEST_F()
86 HloOpcode::kReduce); in TEST_F()
124 HloOpcode::kReduce); in TEST_F()
148 ASSERT_EQ(reduce->fused_expression_root()->opcode(), HloOpcode::kReduce); in TEST_F()
184 HloOpcode::kReduce); in TEST_F()
213 ASSERT_EQ(reduce->opcode(), HloOpcode::kReduce); in TEST_F()
232 ASSERT_EQ(reduce->opcode(), HloOpcode::kReduce); in TEST_F()
250 ASSERT_EQ(reduce->opcode(), HloOpcode::kReduce); in TEST_F()
|
| D | gpu_fusible.cc | 54 if (instr.opcode() == HloOpcode::kReduce && in IfFusedReadsElementsMultipleTimes()
138 if (instr.opcode() == HloOpcode::kReduce && instr.shape().IsTuple()) { in IsInputFusibleReduction()
229 (instr.opcode() == HloOpcode::kReduce && in IsLoopFusible()
321 if (instr.opcode() == HloOpcode::kReduce && in SharedMemoryUsage()
|
| /external/tensorflow/tensorflow/compiler/xla/service/cpu/ |
| D | cpu_instruction_fusion.cc | 38 hlo.opcode() == HloOpcode::kReduce || in CanBeLoopFused()
173 if (consumer->opcode() == HloOpcode::kReduce) { in ShouldFuse()
178 if (producer->opcode() == HloOpcode::kReduce) { in ShouldFuse()
|
| D | parallel_task_assignment.cc | 157 opcode == HloOpcode::kPad || opcode == HloOpcode::kReduce || in GetTargetParallelTaskCount()
|
| /external/tensorflow/tensorflow/compiler/xla/service/ |
| D | while_loop_expensive_invariant_code_motion_test.cc | 95 instr.opcode() == HloOpcode::kReduce; in TEST_F()
179 instr.opcode() == HloOpcode::kReduce; in TEST_F()
|
| D | hlo_opcode_test.cc | 64 case HloOpcode::kReduce: in TEST()
|
| D | bfloat16_support.cc | 114 case HloOpcode::kReduce: in EffectiveOperandPrecisionIsOutputPrecision()
|
| D | fusion_node_indexing_evaluation.cc | 55 case HloOpcode::kReduce: in OpInvalidatesCache()
|
| D | hlo_opcode.h | 120 V(kReduce, "reduce", kHloOpcodeIsVariadic) \
|
| D | hlo_cse.cc | 143 case HloOpcode::kReduce: in CseHash()
|
| D | hlo_element_type_converter.cc | 146 opcode == HloOpcode::kMap || opcode == HloOpcode::kReduce || in Run()
|
| D | bfloat16_normalization_test.cc | 41 hlo.opcode() == HloOpcode::kReduce || in SupportsBF16Operand()
55 if (hlo.opcode() == HloOpcode::kAdd || hlo.opcode() == HloOpcode::kReduce || in SupportsBF16Output()
|
| D | sharding_propagation.cc | 263 case HloOpcode::kReduce: in PickRepresentativeOperand()
321 case HloOpcode::kReduce: in SupportSpatialPartitioning()
592 instruction->opcode() == HloOpcode::kReduce || in InferShardingFromOperands()
683 case HloOpcode::kReduce: { in InferShardingFromOperands()
1309 case HloOpcode::kReduce: { in GetShardingFromUser()
|
| D | hlo_instruction.cc | 256 case HloOpcode::kReduce: in CreateFromProto()
1672 case HloOpcode::kReduce: in CloneWithNewOperands()
2210 case HloOpcode::kReduce: in IdenticalSlowPath()
2433 case HloOpcode::kReduce: in to_apply()
2454 case HloOpcode::kReduce: in set_to_apply()
2824 opcode() == HloOpcode::kReduce || in ExtraAttributesToString()
2883 case HloOpcode::kReduce: in ExtraAttributesToString()
3035 case HloOpcode::kReduce: in IsFusible()
3152 case HloOpcode::kReduce: in Visit()
3591 case HloOpcode::kReduce: in OperandElementUse()
|
| D | hlo_replication_analysis.cc | 143 hlo->opcode() == HloOpcode::kReduce || // in DetermineHloInstructionIsReplicated()
|
| D | call_graph.cc | 63 case HloOpcode::kReduce: in GetInstructionCallContext()
|
| D | conditional_simplifier.cc | 123 case HloOpcode::kReduce: in TryRemoveConditional()
|
| D | instruction_fusion.cc | 172 case HloOpcode::kReduce: in IsExpensive()
|
| D | hlo_module.cc | 171 case HloOpcode::kReduce: in ReplaceComputations()
|
| /external/tensorflow/tensorflow/compiler/xla/tests/ |
| D | reduce_hlo_test.cc | 80 ASSERT_EQ(reduce_instruction->opcode(), HloOpcode::kReduce); in XLA_TEST_P()
|
| D | test_utils.cc | 485 (opcode == HloOpcode::kReduce && in NeedsInitValue()
601 case HloOpcode::kReduce: in CreateLiteralForConstrainedUses()
|
| /external/tensorflow/tensorflow/core/kernels/linalg/ |
| D | einsum_op_impl.h | 85 kReduce = 4, enumerator
100 return kReduce; in GetDimensionType()
512 if (reshape[kReduce] == 1) { // No need to actually reduce. in ReduceOperand()
525 .shaped<T, 2>({output_size, reshape[kReduce]}), in ReduceOperand()
|
| /external/tensorflow/tensorflow/core/nccl/ |
| D | nccl_manager.h | 206 kReduce = 3, enumerator
|
| D | nccl_manager.cc | 460 AddParticipant(std::move(participant), context, kReduce, reduction_op); in AddReduceSend()
467 AddParticipant(std::move(participant), context, kReduce, reduction_op); in AddReduceRecv()
794 case kReduce: { in LoopKernelLaunches()
|
| /external/tensorflow/tensorflow/compiler/mlir/g3doc/ |
| D | xla_gpu_codegen.md | 104 * It's mostly portable and high-level (e.g. unlike GPU kReduce and GPU kCopy).
180 only exception is kReduce.
267 This calls on a rewrite/refactor for kCopy and kReduce. kReduce is already
|