128 lines · plain
1// RUN: mlir-opt %s --linalg-generalize-named-ops --pre-sparsification-rewrite --sparse-reinterpret-map --sparsification --sparse-tensor-conversion --canonicalize --cse | FileCheck %s2 3#DCSR = #sparse_tensor.encoding<{ map = (d0, d1) -> (d0 : compressed, d1 : compressed) }>4 5// CHECK-LABEL: func.func @fill_zero_after_alloc(6// CHECK-SAME: %[[VAL_0:.*]]: !llvm.ptr,7// CHECK-SAME: %[[VAL_1:.*]]: !llvm.ptr) -> !llvm.ptr {8// CHECK-DAG: %[[VAL_2:.*]] = arith.constant 0.000000e+00 : f649// CHECK-DAG: %[[ZERO:.*]] = llvm.mlir.zero : !llvm.ptr10// CHECK-DAG: %[[VAL_3:.*]] = arith.constant 1 : i3211// CHECK-DAG: %[[VAL_4:.*]] = arith.constant 0 : i3212// CHECK-DAG: %[[VAL_5:.*]] = arith.constant 0 : index13// CHECK-DAG: %[[VAL_6:.*]] = arith.constant 1 : index14// CHECK-DAG: %[[VAL_7:.*]] = arith.constant false15// CHECK-DAG: %[[VAL_8:.*]] = arith.constant true16// CHECK-DAG: %[[VAL_9:.*]] = arith.constant 100 : index17// CHECK-DAG: %[[VAL_10:.*]] = arith.constant 300 : index18// CHECK-DAG: %[[VAL_11:.*]] = arith.constant 262144 : i6419// CHECK: %[[VAL_12:.*]] = memref.alloca() : memref<2xi64>20// CHECK: %[[VAL_13:.*]] = memref.cast %[[VAL_12]] : memref<2xi64> to memref<?xi64>21// CHECK: memref.store %[[VAL_11]], %[[VAL_12]]{{\[}}%[[VAL_5]]] : memref<2xi64>22// CHECK: memref.store %[[VAL_11]], %[[VAL_12]]{{\[}}%[[VAL_6]]] : memref<2xi64>23// CHECK: %[[VAL_14:.*]] = memref.alloca() : memref<2xindex>24// CHECK: %[[VAL_15:.*]] = memref.cast %[[VAL_14]] : memref<2xindex> to memref<?xindex>25// CHECK: memref.store %[[VAL_9]], %[[VAL_14]]{{\[}}%[[VAL_5]]] : memref<2xindex>26// CHECK: memref.store %[[VAL_10]], %[[VAL_14]]{{\[}}%[[VAL_6]]] : memref<2xindex>27// CHECK: %[[VAL_16:.*]] = memref.alloca() : memref<2xindex>28// CHECK: %[[VAL_17:.*]] = memref.cast %[[VAL_16]] : memref<2xindex> to memref<?xindex>29// CHECK: memref.store %[[VAL_5]], %[[VAL_16]]{{\[}}%[[VAL_5]]] : memref<2xindex>30// CHECK: memref.store %[[VAL_6]], %[[VAL_16]]{{\[}}%[[VAL_6]]] : memref<2xindex>31// CHECK: %[[VAL_19:.*]] = call @newSparseTensor(%[[VAL_15]], %[[VAL_15]], %[[VAL_13]], %[[VAL_17]], %[[VAL_17]], %[[VAL_4]], %[[VAL_4]], %[[VAL_3]], %[[VAL_4]], %[[ZERO]]) : (memref<?xindex>, memref<?xindex>, memref<?xi64>, memref<?xindex>, memref<?xindex>, i32, i32, i32, i32, !llvm.ptr) -> !llvm.ptr32// CHECK: %[[VAL_20:.*]] = memref.alloc() : memref<300xf64>33// CHECK: %[[VAL_21:.*]] = memref.cast %[[VAL_20]] : memref<300xf64> to memref<?xf64>34// CHECK: %[[VAL_22:.*]] = memref.alloc() : memref<300xi1>35// CHECK: %[[VAL_23:.*]] = memref.cast %[[VAL_22]] : memref<300xi1> to memref<?xi1>36// CHECK: %[[VAL_24:.*]] = memref.alloc() : memref<300xindex>37// CHECK: %[[VAL_25:.*]] = memref.cast %[[VAL_24]] : memref<300xindex> to memref<?xindex>38// CHECK: linalg.fill ins(%[[VAL_2]] : f64) outs(%[[VAL_20]] : memref<300xf64>)39// CHECK: linalg.fill ins(%[[VAL_7]] : i1) outs(%[[VAL_22]] : memref<300xi1>)40// CHECK-DAG: %[[VAL_26:.*]] = call @sparsePositions0(%[[VAL_0]], %[[VAL_5]]) : (!llvm.ptr, index) -> memref<?xindex>41// CHECK-DAG: %[[VAL_27:.*]] = call @sparseCoordinates0(%[[VAL_0]], %[[VAL_5]]) : (!llvm.ptr, index) -> memref<?xindex>42// CHECK-DAG: %[[VAL_28:.*]] = call @sparsePositions0(%[[VAL_0]], %[[VAL_6]]) : (!llvm.ptr, index) -> memref<?xindex>43// CHECK-DAG: %[[VAL_29:.*]] = call @sparseCoordinates0(%[[VAL_0]], %[[VAL_6]]) : (!llvm.ptr, index) -> memref<?xindex>44// CHECK-DAG: %[[VAL_30:.*]] = call @sparseValuesF64(%[[VAL_0]]) : (!llvm.ptr) -> memref<?xf64>45// CHECK-DAG: %[[VAL_31:.*]] = call @sparsePositions0(%[[VAL_1]], %[[VAL_5]]) : (!llvm.ptr, index) -> memref<?xindex>46// CHECK-DAG: %[[VAL_32:.*]] = call @sparseCoordinates0(%[[VAL_1]], %[[VAL_5]]) : (!llvm.ptr, index) -> memref<?xindex>47// CHECK-DAG: %[[VAL_33:.*]] = call @sparsePositions0(%[[VAL_1]], %[[VAL_6]]) : (!llvm.ptr, index) -> memref<?xindex>48// CHECK-DAG: %[[VAL_34:.*]] = call @sparseCoordinates0(%[[VAL_1]], %[[VAL_6]]) : (!llvm.ptr, index) -> memref<?xindex>49// CHECK-DAG: %[[VAL_35:.*]] = call @sparseValuesF64(%[[VAL_1]]) : (!llvm.ptr) -> memref<?xf64>50// CHECK: %[[VAL_36:.*]] = memref.load %[[VAL_26]]{{\[}}%[[VAL_5]]] : memref<?xindex>51// CHECK: %[[VAL_37:.*]] = memref.load %[[VAL_26]]{{\[}}%[[VAL_6]]] : memref<?xindex>52// CHECK: scf.for %[[VAL_38:.*]] = %[[VAL_36]] to %[[VAL_37]] step %[[VAL_6]] {53// CHECK: %[[VAL_39:.*]] = memref.load %[[VAL_27]]{{\[}}%[[VAL_38]]] : memref<?xindex>54// CHECK: %[[VAL_40:.*]] = memref.load %[[VAL_28]]{{\[}}%[[VAL_38]]] : memref<?xindex>55// CHECK: %[[VAL_41:.*]] = arith.addi %[[VAL_38]], %[[VAL_6]] : index56// CHECK: %[[VAL_42:.*]] = memref.load %[[VAL_28]]{{\[}}%[[VAL_41]]] : memref<?xindex>57// CHECK: %[[VAL_43:.*]] = memref.load %[[VAL_31]]{{\[}}%[[VAL_5]]] : memref<?xindex>58// CHECK: %[[VAL_44:.*]] = memref.load %[[VAL_31]]{{\[}}%[[VAL_6]]] : memref<?xindex>59// CHECK: %[[VAL_45:.*]]:3 = scf.while (%[[VAL_46:.*]] = %[[VAL_40]], %[[VAL_47:.*]] = %[[VAL_43]], %[[VAL_48:.*]] = %[[VAL_5]]) : (index, index, index) -> (index, index, index) {60// CHECK: %[[VAL_49:.*]] = arith.cmpi ult, %[[VAL_46]], %[[VAL_42]] : index61// CHECK: %[[VAL_50:.*]] = arith.cmpi ult, %[[VAL_47]], %[[VAL_44]] : index62// CHECK: %[[VAL_51:.*]] = arith.andi %[[VAL_49]], %[[VAL_50]] : i163// CHECK: scf.condition(%[[VAL_51]]) %[[VAL_46]], %[[VAL_47]], %[[VAL_48]] : index, index, index64// CHECK: } do {65// CHECK: ^bb0(%[[VAL_52:.*]]: index, %[[VAL_53:.*]]: index, %[[VAL_54:.*]]: index):66// CHECK: %[[VAL_55:.*]] = memref.load %[[VAL_29]]{{\[}}%[[VAL_52]]] : memref<?xindex>67// CHECK: %[[VAL_56:.*]] = memref.load %[[VAL_32]]{{\[}}%[[VAL_53]]] : memref<?xindex>68// CHECK: %[[VAL_57:.*]] = arith.cmpi ult, %[[VAL_56]], %[[VAL_55]] : index69// CHECK: %[[VAL_58:.*]] = arith.select %[[VAL_57]], %[[VAL_56]], %[[VAL_55]] : index70// CHECK: %[[VAL_59:.*]] = arith.cmpi eq, %[[VAL_55]], %[[VAL_58]] : index71// CHECK: %[[VAL_60:.*]] = arith.cmpi eq, %[[VAL_56]], %[[VAL_58]] : index72// CHECK: %[[VAL_61:.*]] = arith.andi %[[VAL_59]], %[[VAL_60]] : i173// CHECK: %[[VAL_62:.*]] = scf.if %[[VAL_61]] -> (index) {74// CHECK: %[[VAL_63:.*]] = memref.load %[[VAL_30]]{{\[}}%[[VAL_52]]] : memref<?xf64>75// CHECK: %[[VAL_64:.*]] = memref.load %[[VAL_33]]{{\[}}%[[VAL_53]]] : memref<?xindex>76// CHECK: %[[VAL_65:.*]] = arith.addi %[[VAL_53]], %[[VAL_6]] : index77// CHECK: %[[VAL_66:.*]] = memref.load %[[VAL_33]]{{\[}}%[[VAL_65]]] : memref<?xindex>78// CHECK: %[[VAL_67:.*]] = scf.for %[[VAL_68:.*]] = %[[VAL_64]] to %[[VAL_66]] step %[[VAL_6]] iter_args(%[[VAL_69:.*]] = %[[VAL_54]]) -> (index) {79// CHECK: %[[VAL_70:.*]] = memref.load %[[VAL_34]]{{\[}}%[[VAL_68]]] : memref<?xindex>80// CHECK: %[[VAL_71:.*]] = memref.load %[[VAL_20]]{{\[}}%[[VAL_70]]] : memref<300xf64>81// CHECK: %[[VAL_72:.*]] = memref.load %[[VAL_35]]{{\[}}%[[VAL_68]]] : memref<?xf64>82// CHECK: %[[VAL_73:.*]] = arith.mulf %[[VAL_63]], %[[VAL_72]] : f6483// CHECK: %[[VAL_74:.*]] = arith.addf %[[VAL_71]], %[[VAL_73]] : f6484// CHECK: %[[VAL_75:.*]] = memref.load %[[VAL_22]]{{\[}}%[[VAL_70]]] : memref<300xi1>85// CHECK: %[[VAL_76:.*]] = arith.cmpi eq, %[[VAL_75]], %[[VAL_7]] : i186// CHECK: %[[VAL_77:.*]] = scf.if %[[VAL_76]] -> (index) {87// CHECK: memref.store %[[VAL_8]], %[[VAL_22]]{{\[}}%[[VAL_70]]] : memref<300xi1>88// CHECK: memref.store %[[VAL_70]], %[[VAL_24]]{{\[}}%[[VAL_69]]] : memref<300xindex>89// CHECK: %[[VAL_78:.*]] = arith.addi %[[VAL_69]], %[[VAL_6]] : index90// CHECK: scf.yield %[[VAL_78]] : index91// CHECK: } else {92// CHECK: scf.yield %[[VAL_69]] : index93// CHECK: }94// CHECK: memref.store %[[VAL_74]], %[[VAL_20]]{{\[}}%[[VAL_70]]] : memref<300xf64>95// CHECK: scf.yield %[[VAL_77]] : index96// CHECK: }97// CHECK: scf.yield %[[VAL_67]] : index98// CHECK: } else {99// CHECK: scf.yield %[[VAL_54]] : index100// CHECK: }101// CHECK: %[[VAL_79:.*]] = arith.addi %[[VAL_52]], %[[VAL_6]] : index102// CHECK: %[[VAL_80:.*]] = arith.select %[[VAL_59]], %[[VAL_79]], %[[VAL_52]] : index103// CHECK: %[[VAL_81:.*]] = arith.addi %[[VAL_53]], %[[VAL_6]] : index104// CHECK: %[[VAL_82:.*]] = arith.select %[[VAL_60]], %[[VAL_81]], %[[VAL_53]] : index105// CHECK: scf.yield %[[VAL_80]], %[[VAL_82]], %[[VAL_62]] : index, index, index106// CHECK: }107// CHECK: %[[VAL_83:.*]] = memref.alloca() : memref<2xindex>108// CHECK: %[[VAL_84:.*]] = memref.cast %[[VAL_83]] : memref<2xindex> to memref<?xindex>109// CHECK: memref.store %[[VAL_39]], %[[VAL_83]]{{\[}}%[[VAL_5]]] : memref<2xindex>110// CHECK: func.call @expInsertF64(%[[VAL_19]], %[[VAL_84]], %[[VAL_21]], %[[VAL_23]], %[[VAL_25]], %[[VAL_85:.*]]#2) : (!llvm.ptr, memref<?xindex>, memref<?xf64>, memref<?xi1>, memref<?xindex>, index) -> ()111// CHECK: }112// CHECK: memref.dealloc %[[VAL_20]] : memref<300xf64>113// CHECK: memref.dealloc %[[VAL_22]] : memref<300xi1>114// CHECK: memref.dealloc %[[VAL_24]] : memref<300xindex>115// CHECK: call @endLexInsert(%[[VAL_19]]) : (!llvm.ptr) -> ()116// CHECK: return %[[VAL_19]] : !llvm.ptr117// CHECK: }118func.func @fill_zero_after_alloc(%arg0: tensor<100x200xf64, #DCSR>,119 %arg1: tensor<200x300xf64, #DCSR>) -> tensor<100x300xf64, #DCSR> {120 %0 = tensor.empty() : tensor<100x300xf64, #DCSR>121 %cst = arith.constant 0.000000e+00 : f64122 %1 = linalg.fill ins(%cst : f64)123 outs(%0 : tensor<100x300xf64, #DCSR>) -> tensor<100x300xf64, #DCSR>124 %2 = linalg.matmul ins(%arg0, %arg1 : tensor<100x200xf64, #DCSR>, tensor<200x300xf64, #DCSR>)125 outs(%1 : tensor<100x300xf64, #DCSR>) -> tensor<100x300xf64, #DCSR>126 return %2 : tensor<100x300xf64, #DCSR>127}128