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 direct IR generation and 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 direct IR generation and VLA vectorization. 32// RUN: %if mlir_arm_sve_tests %{ %{compile_sve} | %{run_sve} | FileCheck %s %} 33 34#SparseVector = #sparse_tensor.encoding<{map = (d0) -> (d0 : compressed)}> 35 36#trait_op = { 37 indexing_maps = [ 38 affine_map<(i) -> (i)>, // a (in) 39 affine_map<(i) -> (i)> // x (out) 40 ], 41 iterator_types = ["parallel"], 42 doc = "x(i) = OP a(i)" 43} 44 45module { 46 func.func @cre(%arga: tensor<?xcomplex<f32>, #SparseVector>) 47 -> tensor<?xf32, #SparseVector> { 48 %c = arith.constant 0 : index 49 %d = tensor.dim %arga, %c : tensor<?xcomplex<f32>, #SparseVector> 50 %xv = tensor.empty(%d) : tensor<?xf32, #SparseVector> 51 %0 = linalg.generic #trait_op 52 ins(%arga: tensor<?xcomplex<f32>, #SparseVector>) 53 outs(%xv: tensor<?xf32, #SparseVector>) { 54 ^bb(%a: complex<f32>, %x: f32): 55 %1 = complex.re %a : complex<f32> 56 linalg.yield %1 : f32 57 } -> tensor<?xf32, #SparseVector> 58 return %0 : tensor<?xf32, #SparseVector> 59 } 60 61 func.func @cim(%arga: tensor<?xcomplex<f32>, #SparseVector>) 62 -> tensor<?xf32, #SparseVector> { 63 %c = arith.constant 0 : index 64 %d = tensor.dim %arga, %c : tensor<?xcomplex<f32>, #SparseVector> 65 %xv = tensor.empty(%d) : tensor<?xf32, #SparseVector> 66 %0 = linalg.generic #trait_op 67 ins(%arga: tensor<?xcomplex<f32>, #SparseVector>) 68 outs(%xv: tensor<?xf32, #SparseVector>) { 69 ^bb(%a: complex<f32>, %x: f32): 70 %1 = complex.im %a : complex<f32> 71 linalg.yield %1 : f32 72 } -> tensor<?xf32, #SparseVector> 73 return %0 : tensor<?xf32, #SparseVector> 74 } 75 76 func.func @main() { 77 // Setup sparse vectors. 78 %v1 = arith.constant sparse< 79 [ [0], [20], [31] ], 80 [ (5.13, 2.0), (3.0, 4.0), (5.0, 6.0) ] > : tensor<32xcomplex<f32>> 81 %sv1 = sparse_tensor.convert %v1 : tensor<32xcomplex<f32>> to tensor<?xcomplex<f32>, #SparseVector> 82 83 // Call sparse vector kernels. 84 %0 = call @cre(%sv1) 85 : (tensor<?xcomplex<f32>, #SparseVector>) -> tensor<?xf32, #SparseVector> 86 87 %1 = call @cim(%sv1) 88 : (tensor<?xcomplex<f32>, #SparseVector>) -> tensor<?xf32, #SparseVector> 89 90 // 91 // Verify the results. 92 // 93 // CHECK: ---- Sparse Tensor ---- 94 // CHECK-NEXT: nse = 3 95 // CHECK-NEXT: dim = ( 32 ) 96 // CHECK-NEXT: lvl = ( 32 ) 97 // CHECK-NEXT: pos[0] : ( 0, 3 ) 98 // CHECK-NEXT: crd[0] : ( 0, 20, 31 ) 99 // CHECK-NEXT: values : ( 5.13, 3, 5 ) 100 // CHECK-NEXT: ---- 101 // 102 // CHECK-NEXT: ---- Sparse Tensor ---- 103 // CHECK-NEXT: nse = 3 104 // CHECK-NEXT: dim = ( 32 ) 105 // CHECK-NEXT: lvl = ( 32 ) 106 // CHECK-NEXT: pos[0] : ( 0, 3 ) 107 // CHECK-NEXT: crd[0] : ( 0, 20, 31 ) 108 // CHECK-NEXT: values : ( 2, 4, 6 ) 109 // CHECK-NEXT: ---- 110 // 111 sparse_tensor.print %0 : tensor<?xf32, #SparseVector> 112 sparse_tensor.print %1 : tensor<?xf32, #SparseVector> 113 114 // Release the resources. 115 bufferization.dealloc_tensor %sv1 : tensor<?xcomplex<f32>, #SparseVector> 116 bufferization.dealloc_tensor %0 : tensor<?xf32, #SparseVector> 117 bufferization.dealloc_tensor %1 : tensor<?xf32, #SparseVector> 118 return 119 } 120} 121