/external/tensorflow/tensorflow/compiler/xla/mlir_hlo/stablehlo/tests/ |
D | verify_reduce_window.mlir | 3 // CHECK: func @reduce_window 4 func.func @reduce_window(%arg0: tensor<4x2xf32>, %arg1: tensor<4x2xi32>, 7 %0:2 = "stablehlo.reduce_window"(%arg0, %arg1, %init0, %init1) ({ 25 %0:2 = "stablehlo.reduce_window"(%arg0, %arg1, %init0, %init1) ({ 44 %0 = "stablehlo.reduce_window"(%arg0, %init0) ({ 60 %0 = "stablehlo.reduce_window"(%arg0, %init0) ({ 78 %0:2 = "stablehlo.reduce_window"() ({ 98 %0:2 = "stablehlo.reduce_window"(%arg0, %arg1, %init0, %init1) ({ 119 %0:2 = "stablehlo.reduce_window"(%arg0, %arg1, %init0, %init1) ({ 140 %0:2 = "stablehlo.reduce_window"(%arg0, %arg1, %init0, %init1) ({ [all …]
|
/external/tensorflow/tensorflow/compiler/xla/mlir_hlo/tests/Dialect/mhlo/ |
D | verifier_reduce_window_op.mlir | 3 // CHECK: func @reduce_window 4 func.func @reduce_window(%arg0: tensor<4x2xf32>, %arg1: tensor<4x2xi32>, 7 %0:2 = "mhlo.reduce_window"(%arg0, %arg1, %init0, %init1) ({ 25 %0:2 = "mhlo.reduce_window"(%arg0, %arg1, %init0, %init1) ({ 44 %0 = "mhlo.reduce_window"(%arg0, %init0) ({ 60 %0 = "mhlo.reduce_window"(%arg0, %init0) ({ 78 %0:2 = "mhlo.reduce_window"() ({ 98 %0:2 = "mhlo.reduce_window"(%arg0, %arg1, %init0, %init1) ({ 119 %0:2 = "mhlo.reduce_window"(%arg0, %arg1, %init0, %init1) ({ 140 %0:2 = "mhlo.reduce_window"(%arg0, %arg1, %init0, %init1) ({ [all …]
|
D | hlo-legalize-to-lhlo.mlir | 557 // CHECK-LABEL: func @reduce_window 558 func.func @reduce_window(%arg0: tensor<1x17x17x64xf32>, %arg1: tensor<f32>) -> tensor<1x8x8x64xf32>… 560 // CHECK: "lmhlo.reduce_window"(%{{.+}}, %{{.+}}, %[[OUT]]) ({ 569 %0 = "mhlo.reduce_window"(%arg0, %arg1) ( { 585 // CHECK: "lmhlo.reduce_window"(%{{.+}}, %{{.+}}, %{{.+}}, %{{.+}}, %[[OUT_F]], %[[OUT_I]]) ({ 597 %0:2 = "mhlo.reduce_window"(%arg0, %arg1, %arg2, %arg3) ( {
|
D | hlo-legalize-to-linalg.mlir | 3409 %0 = "mhlo.reduce_window"(%arg0, %arg1) ({ 3436 %0 = "mhlo.reduce_window"(%arg0, %arg1) ({ 3461 %0 = "mhlo.reduce_window"(%arg0, %arg1) ({ 3486 %1 = "mhlo.reduce_window"(%arg0, %0) ({ 3511 %0:2 = "mhlo.reduce_window"(%arg0, %arg0, %arg1, %arg1) ({ 3548 %0 = "mhlo.reduce_window"(%arg0, %arg1) ({ 3587 %0 = "mhlo.reduce_window"(%arg0, %arg1) ({ 3612 %0 = "mhlo.reduce_window"(%arg0, %arg1) ({ 3637 %0 = "mhlo.reduce_window"(%arg0, %arg1) ({ 3665 %0 = "mhlo.reduce_window"(%arg0, %arg1) ({ [all …]
|
/external/tensorflow/tensorflow/compiler/xla/service/ |
D | tree_reduction_rewriter.cc | 97 HloInstruction *reduce_window = in HandleReduce() local 103 HloInstruction::CreateReduce(reduce_shape, reduce_window, initial_value, in HandleReduce()
|
D | select_and_scatter_expander.cc | 142 auto* reduce_window = in ExpandInstruction() local 160 HloInstruction::CreateGetTupleElement(reduce_window, i)); in ExpandInstruction()
|
D | algebraic_simplifier.cc | 5604 auto* reduce_window = Cast<HloReduceWindowInstruction>(hlo); in HandleReduceWindow() local 5605 const bool multi_output_reduce_window = reduce_window->shape().IsTuple(); in HandleReduceWindow() 5606 auto inputs = reduce_window->inputs(); in HandleReduceWindow() 5607 auto init_values = reduce_window->init_values(); in HandleReduceWindow() 5608 auto input_count = reduce_window->input_count(); in HandleReduceWindow() 5609 auto input_shapes = reduce_window->input_shapes(); in HandleReduceWindow() 5610 auto output_shapes = reduce_window->output_shapes(); in HandleReduceWindow() 5614 return ReplaceWithNewInstruction(reduce_window, in HandleReduceWindow() 5617 return ReplaceInstruction(reduce_window, elements[0]); in HandleReduceWindow() 5634 reduce_window->to_apply()->root_instruction()->opcode() == in HandleReduceWindow() [all …]
|
D | hlo_cost_analysis.cc | 410 const HloInstruction* reduce_window) { in HandleReduceWindow() argument 411 const Window& window = reduce_window->window(); in HandleReduceWindow() 412 auto function = reduce_window->to_apply(); in HandleReduceWindow() 425 ShapeUtil::ElementsIn(reduce_window->shape().IsArray() in HandleReduceWindow() 426 ? reduce_window->shape() in HandleReduceWindow() 427 : reduce_window->shape().tuple_shapes(0)); in HandleReduceWindow()
|
D | sharding_propagation.cc | 1419 auto* reduce_window = Cast<HloReduceWindowInstruction>(&user); in GetShardingFromUser() local 1420 if (!absl::c_linear_search(reduce_window->inputs(), &instruction)) { in GetShardingFromUser() 1423 if (reduce_window->shape().IsTuple()) { in GetShardingFromUser() 1424 auto sub_sharding = reduce_window->sharding().GetSubSharding( in GetShardingFromUser() 1425 reduce_window->shape(), in GetShardingFromUser() 1426 {reduce_window->operand_index(&instruction)}); in GetShardingFromUser() 1429 return reduce_window->sharding(); in GetShardingFromUser() 1997 auto* reduce_window = Cast<HloReduceWindowInstruction>(instruction); in InferShardingFromOperands() local 2005 << reduce_window->ToString(); in InferShardingFromOperands() 2009 for (HloInstruction* operand : reduce_window->inputs()) { in InferShardingFromOperands() [all …]
|
D | dfs_hlo_visitor_with_default.h | 199 Status HandleReduceWindow(HloInstructionPtr reduce_window) override { in HandleReduceWindow() argument 200 return DefaultAction(reduce_window); in HandleReduceWindow()
|
D | convolution_group_converter.cc | 402 auto reduce_window = add(HloInstruction::CreateReduceWindow( in HandleBatchGroupCount() local 406 Shape convert_back_shape = reduce_window->shape(); in HandleBatchGroupCount() 411 HloInstruction::CreateConvert(convert_back_shape, reduce_window); in HandleBatchGroupCount()
|
D | hlo_verifier.cc | 1262 Status ShapeVerifier::HandleReduceWindow(HloInstruction* reduce_window) { in HandleReduceWindow() argument 1263 VLOG(2) << "Verify reduce window:" << reduce_window->ToString() << "\n"; in HandleReduceWindow() 1264 auto reduce_window_instr = Cast<HloReduceWindowInstruction>(reduce_window); in HandleReduceWindow() 1268 VLOG(2) << "reduce instruction is :" << reduce_window->ToString() << "\n"; in HandleReduceWindow() 1270 reduce_window, ShapeInference::InferReduceWindowShape( in HandleReduceWindow() 1271 input_shapes, init_shapes, reduce_window->window(), in HandleReduceWindow() 1272 reduce_window->to_apply()->ComputeProgramShape()))); in HandleReduceWindow() 1277 *reduce_window, reduce_window->operand_count()); in HandleReduceWindow()
|
D | dynamic_dimension_inference.cc | 1236 auto* reduce_window = Cast<HloReduceWindowInstruction>(hlo); in HandleReduceWindow() local 1238 reduce_window->window().dimensions(dimension); in HandleReduceWindow() 1240 if (operand_index >= reduce_window->input_count()) { in HandleReduceWindow() 1257 reduce_window->shape(), in HandleReduceWindow() 1259 if (!ShapeUtil::IsLeafIndex(reduce_window->shape(), in HandleReduceWindow() 1263 parent_->SetDynamicSize(reduce_window, reduce_window_result_index, in HandleReduceWindow()
|
D | sharding_propagation_test.cc | 1526 auto* reduce_window = FindInstruction(module.get(), "reduce-window"); in TEST_P() local 1527 ASSERT_NE(reduce_window, nullptr); in TEST_P() 1528 EXPECT_THAT(reduce_window, op::Sharding("{devices=[2,1]0,1}")); in TEST_P() 1532 EXPECT_THAT(reduce_window->sharding(), in TEST_P() 1536 EXPECT_THAT(reduce_window->sharding(), ShardingMetadata({})); in TEST_P() 1579 auto* reduce_window = FindInstruction(module.get(), "reduce-window"); in TEST_P() local 1580 ASSERT_NE(reduce_window, nullptr); in TEST_P() 1581 EXPECT_THAT(reduce_window, in TEST_P() 1588 EXPECT_THAT(reduce_window->sharding().tuple_elements()[0], in TEST_P() 1590 EXPECT_THAT(reduce_window->sharding().tuple_elements()[1], in TEST_P() [all …]
|
D | hlo_evaluator_typed_visitor.h | 1758 Status HandleReduceWindow(HloInstruction* reduce_window) override { 1759 auto* reduce_window_instr = Cast<HloReduceWindowInstruction>(reduce_window); 1760 const Window& window = reduce_window->window(); 1761 HloComputation* function = reduce_window->to_apply(); 1769 ShapeUtil::Compatible(reduce_window->shape(), inferred_return_shape)) 1771 << ShapeUtil::HumanStringWithLayout(reduce_window->shape()) 1900 parent_->evaluated_[reduce_window] = std::move(result);
|
D | hlo_cost_analysis.h | 154 Status HandleReduceWindow(const HloInstruction* reduce_window) override;
|
D | elemental_ir_emitter.cc | 2784 const HloReduceWindowInstruction* reduce_window, in EmitElementalReduceWindow() argument 2797 int64_t input_count = reduce_window->input_count(); in EmitElementalReduceWindow() 2803 auto operand = reduce_window->inputs()[operand_index]; in EmitElementalReduceWindow() 2827 const Window& window = reduce_window->window(); in EmitElementalReduceWindow() 2828 llvm_ir::ForLoopNest loops(IrName(reduce_window), b_, index_type); in EmitElementalReduceWindow() 2874 reduce_window->inputs()[0]->shape().dimensions(i)))); in EmitElementalReduceWindow() 2882 std::vector<llvm::Value*> input_values(reduce_window->operand_count()); in EmitElementalReduceWindow() 2884 reduce_window->inputs()[0]->shape(), index_type); in EmitElementalReduceWindow() 2895 EmitThreadLocalCall(*reduce_window->to_apply(), in EmitElementalReduceWindow() 2906 reduce_window->shape().IsTuple()); in EmitElementalReduceWindow()
|
D | hlo_verifier.h | 195 Status HandleReduceWindow(HloInstruction* reduce_window) override;
|
D | elemental_ir_emitter.h | 281 const HloReduceWindowInstruction* reduce_window,
|
/external/tensorflow/tensorflow/compiler/tests/ |
D | reduce_window_test.py | 34 output = xla.reduce_window(placeholder, init, reducer, **kwargs)
|
/external/tensorflow/tensorflow/compiler/xla/mlir_hlo/tests/Dialect/lhlo/ |
D | lhlo-legalize-to-parallel-loops.mlir | 135 func.func @reduce_window(%arg: memref<112x112xf32>, 138 "lmhlo.reduce_window"(%arg, %init, %result) ({ 150 // CHECK-LABEL: func @reduce_window(
|
/external/tensorflow/tensorflow/compiler/mlir/xla/tests/ |
D | legalize-tf-no-tf2xla-fallback.mlir | 1521 // CHECK: "mhlo.reduce_window"(%[[ARG]], %[[INIT]]) 1547 // CHECK: "mhlo.reduce_window"(%[[ARG]], %[[INIT]]) 5210 // CHECK: [[DIVIDEND:%.+]] = "mhlo.reduce_window"([[CONV32]], [[ZERO]]) ({ 5236 // CHECK: [[DIVIDEND:%.+]] = "mhlo.reduce_window"([[CONV32]], [[ZERO]]) ({ 5262 // CHECK: [[DIVIDEND:%.+]] = "mhlo.reduce_window"([[CONV32]], [[ZERO]]) ({ 5288 // CHECK: [[DIVIDEND:%.+]] = "mhlo.reduce_window"([[CONV32]], [[ZERO]]) ({ 5313 // CHECK: %[[DIVIDEND:.*]] = "mhlo.reduce_window"(%[[ARG0]], %[[ZERO]]) ({ 5323 // CHECK: %[[DIVISOR:.*]] = "mhlo.reduce_window"(%[[ONES]], %[[ZERO]]) ({ 5345 // CHECK: %[[DIVIDEND:.*]] = "mhlo.reduce_window"(%[[ARG0]], %[[ZERO]]) ({ 5355 // CHECK: %[[DIVISOR:.*]] = "mhlo.reduce_window"(%[[ONES]], %[[ZERO]]) ({ [all …]
|
D | legalize-tf.mlir | 1538 // CHECK: "mhlo.reduce_window"(%[[ARG]], %[[INIT]]) 1564 // CHECK: "mhlo.reduce_window"(%[[ARG]], %[[INIT]]) 5378 // CHECK: [[DIVIDEND:%.+]] = "mhlo.reduce_window"([[CONV32]], [[ZERO]]) ({ 5404 // CHECK: [[DIVIDEND:%.+]] = "mhlo.reduce_window"([[CONV32]], [[ZERO]]) ({ 5430 // CHECK: [[DIVIDEND:%.+]] = "mhlo.reduce_window"([[CONV32]], [[ZERO]]) ({ 5456 // CHECK: [[DIVIDEND:%.+]] = "mhlo.reduce_window"([[CONV32]], [[ZERO]]) ({ 5481 // CHECK: %[[DIVIDEND:.*]] = "mhlo.reduce_window"(%[[ARG0]], %[[ZERO]]) ({ 5491 // CHECK: %[[DIVISOR:.*]] = "mhlo.reduce_window"(%[[ONES]], %[[ZERO]]) ({ 5513 // CHECK: %[[DIVIDEND:.*]] = "mhlo.reduce_window"(%[[ARG0]], %[[ZERO]]) ({ 5523 // CHECK: %[[DIVISOR:.*]] = "mhlo.reduce_window"(%[[ONES]], %[[ZERO]]) ({ [all …]
|
/external/tensorflow/tensorflow/compiler/xla/service/cpu/ |
D | ir_emitter.h | 157 Status HandleReduceWindow(HloInstruction* reduce_window) override;
|
/external/tensorflow/tensorflow/compiler/tf2xla/python/ |
D | xla.py | 416 def reduce_window(operand, function
|