1//-------------------------------------------------------------------------------------------------- 2// WHEN CREATING A NEW TEST, PLEASE JUST COPY & PASTE WITHOUT EDITS. 3// 4// Set-up that's shared across all tests in this directory. In principle, this 5// config could be moved to lit.local.cfg. However, there are downstream users that 6// do not use these LIT config files. Hence why this is kept inline. 7// 8// DEFINE: %{sparsifier_opts} = enable-runtime-library=true 9// DEFINE: %{sparsifier_opts_sve} = enable-arm-sve=true %{sparsifier_opts} 10// DEFINE: %{compile} = mlir-opt %s --sparsifier="%{sparsifier_opts}" 11// DEFINE: %{compile_sve} = mlir-opt %s --sparsifier="%{sparsifier_opts_sve}" 12// DEFINE: %{run_libs} = -shared-libs=%mlir_c_runner_utils,%mlir_runner_utils 13// DEFINE: %{run_libs_sve} = -shared-libs=%native_mlir_runner_utils,%native_mlir_c_runner_utils 14// DEFINE: %{run_opts} = -e main -entry-point-result=void 15// DEFINE: %{run} = mlir-runner %{run_opts} %{run_libs} 16// DEFINE: %{run_sve} = %mcr_aarch64_cmd --march=aarch64 --mattr="+sve" %{run_opts} %{run_libs_sve} 17// 18// DEFINE: %{env} = 19//-------------------------------------------------------------------------------------------------- 20 21// RUN: %{compile} | %{run} | FileCheck %s 22// 23// Do the same run, but now with direct IR generation. 24// REDEFINE: %{sparsifier_opts} = enable-runtime-library=false 25// RUN: %{compile} | %{run} | FileCheck %s 26// 27// Do the same run, but now with vectorization. 28// REDEFINE: %{sparsifier_opts} = enable-runtime-library=false vl=2 reassociate-fp-reductions=true enable-index-optimizations=true 29// RUN: %{compile} | %{run} | FileCheck %s 30// 31// Do the same run, but now with VLA vectorization. 32// RUN: %if mlir_arm_sve_tests %{ %{compile_sve} | %{run_sve} | FileCheck %s %} 33 34#ST1 = #sparse_tensor.encoding<{map = (d0, d1, d2) -> (d0 : compressed, d1 : compressed, d2 : compressed)}> 35#ST2 = #sparse_tensor.encoding<{map = (d0, d1, d2) -> (d0 : compressed, d1 : compressed, d2 : dense)}> 36 37// 38// Trait for 3-d tensor operation. 39// 40#trait_scale = { 41 indexing_maps = [ 42 affine_map<(i,j,k) -> (i,j,k)>, // A (in) 43 affine_map<(i,j,k) -> (i,j,k)> // X (out) 44 ], 45 iterator_types = ["parallel", "parallel", "parallel"], 46 doc = "X(i,j,k) = A(i,j,k) * 2.0" 47} 48 49module { 50 // Scales a sparse tensor into a new sparse tensor. 51 func.func @tensor_scale(%arga: tensor<?x?x?xf64, #ST1>) -> tensor<?x?x?xf64, #ST2> { 52 %s = arith.constant 2.0 : f64 53 %c0 = arith.constant 0 : index 54 %c1 = arith.constant 1 : index 55 %c2 = arith.constant 2 : index 56 %d0 = tensor.dim %arga, %c0 : tensor<?x?x?xf64, #ST1> 57 %d1 = tensor.dim %arga, %c1 : tensor<?x?x?xf64, #ST1> 58 %d2 = tensor.dim %arga, %c2 : tensor<?x?x?xf64, #ST1> 59 %xm = tensor.empty(%d0, %d1, %d2) : tensor<?x?x?xf64, #ST2> 60 %0 = linalg.generic #trait_scale 61 ins(%arga: tensor<?x?x?xf64, #ST1>) 62 outs(%xm: tensor<?x?x?xf64, #ST2>) { 63 ^bb(%a: f64, %x: f64): 64 %1 = arith.mulf %a, %s : f64 65 linalg.yield %1 : f64 66 } -> tensor<?x?x?xf64, #ST2> 67 return %0 : tensor<?x?x?xf64, #ST2> 68 } 69 70 // Driver method to call and verify tensor kernel. 71 func.func @main() { 72 %c0 = arith.constant 0 : index 73 %d1 = arith.constant -1.0 : f64 74 75 // Setup sparse tensor. 76 %t = arith.constant dense< 77 [ [ [1.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ], 78 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ], 79 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ], 80 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 2.0 ] ], 81 [ [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ], 82 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ], 83 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ], 84 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ] ], 85 [ [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ], 86 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ], 87 [0.0, 3.0, 4.0, 0.0, 0.0, 0.0, 0.0, 5.0 ], 88 [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0 ] ] ]> : tensor<3x4x8xf64> 89 %st = sparse_tensor.convert %t : tensor<3x4x8xf64> to tensor<?x?x?xf64, #ST1> 90 91 // Call sparse vector kernels. 92 %0 = call @tensor_scale(%st) : (tensor<?x?x?xf64, #ST1>) -> tensor<?x?x?xf64, #ST2> 93 94 // 95 // Sanity check on stored values. 96 // 97 // CHECK: ---- Sparse Tensor ---- 98 // CHECK-NEXT: nse = 5 99 // CHECK-NEXT: dim = ( 3, 4, 8 ) 100 // CHECK-NEXT: lvl = ( 3, 4, 8 ) 101 // CHECK-NEXT: pos[0] : ( 0, 2 ) 102 // CHECK-NEXT: crd[0] : ( 0, 2 ) 103 // CHECK-NEXT: pos[1] : ( 0, 2, 3 ) 104 // CHECK-NEXT: crd[1] : ( 0, 3, 2 ) 105 // CHECK-NEXT: pos[2] : ( 0, 1, 2, 5 ) 106 // CHECK-NEXT: crd[2] : ( 0, 7, 1, 2, 7 ) 107 // CHECK-NEXT: values : ( 1, 2, 3, 4, 5 ) 108 // CHECK-NEXT: ---- 109 // CHECK: ---- Sparse Tensor ---- 110 // CHECK-NEXT: nse = 24 111 // CHECK-NEXT: dim = ( 3, 4, 8 ) 112 // CHECK-NEXT: lvl = ( 3, 4, 8 ) 113 // CHECK-NEXT: pos[0] : ( 0, 2 ) 114 // CHECK-NEXT: crd[0] : ( 0, 2 ) 115 // CHECK-NEXT: pos[1] : ( 0, 2, 3 ) 116 // CHECK-NEXT: crd[1] : ( 0, 3, 2 ) 117 // CHECK-NEXT: values : ( 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 4, 0, 6, 8, 0, 0, 0, 0, 10 ) 118 // CHECK-NEXT: ---- 119 // 120 sparse_tensor.print %st : tensor<?x?x?xf64, #ST1> 121 sparse_tensor.print %0 : tensor<?x?x?xf64, #ST2> 122 123 // Release the resources. 124 bufferization.dealloc_tensor %st : tensor<?x?x?xf64, #ST1> 125 bufferization.dealloc_tensor %0 : tensor<?x?x?xf64, #ST2> 126 return 127 } 128} 129