This is an archive of the discontinued LLVM Phabricator instance.

Differential D137850

[mlir][sparse] Extend more integration to run on the codegen path.
ClosedPublic

Authored by bixia on Nov 11 2022, 8:50 AM.

Details

Reviewers
aartbik
nicolasvasilache
Peiming
Commits
rGda749560e11b: [mlir][sparse] Extend more integration to run on the codegen path.

Diff Detail

Event Timeline

bixia created this revision.Nov 11 2022, 8:50 AM
Herald added a reviewer: aartbik. · View Herald TranscriptNov 11 2022, 8:50 AM
Herald added a project: Restricted Project. · View Herald Transcript
Herald added subscribers: Moerafaat, anlunx, bzcheeseman and 21 others. · View Herald Transcript
bixia requested review of this revision.Nov 11 2022, 8:50 AM
Herald added a reviewer: nicolasvasilache. · View Herald TranscriptNov 11 2022, 8:50 AM
Herald added a project: Restricted Project. · View Herald Transcript
Herald added subscribers: stephenneuendorffer, nicolasvasilache. · View Herald Transcript
Harbormaster completed remote builds in B197247: Diff 474781.Nov 11 2022, 9:10 AM
bixia added a reviewer: Peiming.Nov 11 2022, 9:12 AM
Peiming accepted this revision.Nov 11 2022, 9:20 AM
This revision is now accepted and ready to land.Nov 11 2022, 9:20 AM
Closed by commit rGda749560e11b: [mlir][sparse] Extend more integration to run on the codegen path. (authored by bixia). · Explain WhyNov 11 2022, 9:42 AM
This revision was automatically updated to reflect the committed changes.
bixia added a commit: rGda749560e11b: [mlir][sparse] Extend more integration to run on the codegen path..
aartbik added inline comments.Nov 11 2022, 1:29 PM
mlir/test/Integration/Dialect/SparseTensor/CPU/dense_output.mlir
7

please add:


 Do the same run, but now with direct IR generation.
//

mlir/test/Integration/Dialect/SparseTensor/CPU/dense_output_bf16.mlir
7

and here


 Do the same run, but now with direct IR generation.
//

mlir/test/Integration/Dialect/SparseTensor/CPU/sparse_conversion_dyn.mlir
60

alternatively, we could simply print one less, and check the noe field for added safety

Revision Contents

Diff 474794

mlir/test/Integration/Dialect/SparseTensor/CPU/dense_output.mlir

// RUN: mlir-opt %s --sparse-compiler | \ // RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=true | \
// RUN: TENSOR0="%mlir_src_dir/test/Integration/data/test.mtx" \
// RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s
aartbikUnsubmitted
Not Done
ReplyInline Actions

please add:


 Do the same run, but now with direct IR generation.
//

aartbik: please add: // // Do the same run, but now with direct IR generation. //
// RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=false | \
// RUN: TENSOR0="%mlir_src_dir/test/Integration/data/test.mtx" \ // RUN: TENSOR0="%mlir_src_dir/test/Integration/data/test.mtx" \
// RUN: mlir-cpu-runner \ // RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \ // RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \ // RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s // RUN: FileCheck %s
!Filename = !llvm.ptr<i8> !Filename = !llvm.ptr<i8>
▲ Show 20 Lines • Show All 87 Lines • Show Last 20 Lines

mlir/test/Integration/Dialect/SparseTensor/CPU/dense_output_bf16.mlir

// RUN: mlir-opt %s --sparse-compiler | \ // RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=true | \
// RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s
// RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=false | \
aartbikUnsubmitted
Not Done
ReplyInline Actions

and here


 Do the same run, but now with direct IR generation.
//

aartbik: and here // // Do the same run, but now with direct IR generation. //
// RUN: mlir-cpu-runner \ // RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \ // RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \ // RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s // RUN: FileCheck %s
// UNSUPPORTED: aarch64 // UNSUPPORTED: aarch64
#SparseVector = #sparse_tensor.encoding<{dimLevelType = ["compressed"]}> #SparseVector = #sparse_tensor.encoding<{dimLevelType = ["compressed"]}>
▲ Show 20 Lines • Show All 83 Lines • Show Last 20 Lines

mlir/test/Integration/Dialect/SparseTensor/CPU/dense_output_f16.mlir

// RUN: mlir-opt %s --sparse-compiler | \ // RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=true | \
// RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s
// RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=false | \
// RUN: mlir-cpu-runner \ // RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \ // RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \ // RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s // RUN: FileCheck %s
#SparseVector = #sparse_tensor.encoding<{dimLevelType = ["compressed"]}> #SparseVector = #sparse_tensor.encoding<{dimLevelType = ["compressed"]}>
#DenseVector = #sparse_tensor.encoding<{dimLevelType = ["dense"]}> #DenseVector = #sparse_tensor.encoding<{dimLevelType = ["dense"]}>
▲ Show 20 Lines • Show All 81 Lines • Show Last 20 Lines

mlir/test/Integration/Dialect/SparseTensor/CPU/sparse_codegen_foreach.mlir

// RUN: mlir-opt %s --sparse-compiler | \ // RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=true | \
// RUN: mlir-cpu-runner -e entry -entry-point-result=void \ // RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s
// RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=false | \
// RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \ // RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s // RUN: FileCheck %s
#Row = #sparse_tensor.encoding<{ #Row = #sparse_tensor.encoding<{
dimLevelType = [ "compressed", "dense" ] dimLevelType = [ "compressed", "dense" ]
}> }>
#CSR = #sparse_tensor.encoding<{ #CSR = #sparse_tensor.encoding<{
▲ Show 20 Lines • Show All 207 Lines • Show Last 20 Lines

mlir/test/Integration/Dialect/SparseTensor/CPU/sparse_conversion_dyn.mlir

// RUN: mlir-opt %s --sparse-compiler | \ // RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=true | \
// RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s
// RUN: mlir-opt %s --sparse-compiler="enable-runtime-library=false enable-buffer-initialization=true" | \
// RUN: mlir-cpu-runner \ // RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \ // RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \ // RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s // RUN: FileCheck %s
#DCSR = #sparse_tensor.encoding<{ #DCSR = #sparse_tensor.encoding<{
dimLevelType = [ "compressed", "compressed" ] dimLevelType = [ "compressed", "compressed" ]
}> }>
Show All 11 Lines
module { module {
// //
// Helper method to print values array. The transfer actually // Helper method to print values array. The transfer actually
// reads more than required to verify size of buffer as well. // reads more than required to verify size of buffer as well.
// //
func.func @dump(%arg0: memref<?xf64>) { func.func @dump(%arg0: memref<?xf64>) {
%c = arith.constant 0 : index %c = arith.constant 0 : index
%d = arith.constant -1.0 : f64 %d = arith.constant 0.0 : f64
%0 = vector.transfer_read %arg0[%c], %d: memref<?xf64>, vector<8xf64> %0 = vector.transfer_read %arg0[%c], %d: memref<?xf64>, vector<8xf64>
vector.print %0 : vector<8xf64> vector.print %0 : vector<8xf64>
return return
} }
func.func @entry() { func.func @entry() {
%t1 = arith.constant sparse< %t1 = arith.constant sparse<
[ [0,0], [0,1], [0,63], [1,0], [1,1], [31,0], [31,63] ], [ [0,0], [0,1], [0,63], [1,0], [1,1], [31,0], [31,63] ],
[ 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0 ]> : tensor<32x64xf64> [ 1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0 ]> : tensor<32x64xf64>
%t2 = tensor.cast %t1 : tensor<32x64xf64> to tensor<?x?xf64> %t2 = tensor.cast %t1 : tensor<32x64xf64> to tensor<?x?xf64>
// Four dense to sparse conversions. // Four dense to sparse conversions.
%1 = sparse_tensor.convert %t1 : tensor<32x64xf64> to tensor<?x?xf64, #DCSR> %1 = sparse_tensor.convert %t1 : tensor<32x64xf64> to tensor<?x?xf64, #DCSR>
%2 = sparse_tensor.convert %t1 : tensor<32x64xf64> to tensor<?x?xf64, #DCSC> %2 = sparse_tensor.convert %t1 : tensor<32x64xf64> to tensor<?x?xf64, #DCSC>
%3 = sparse_tensor.convert %t2 : tensor<?x?xf64> to tensor<?x?xf64, #DCSR> %3 = sparse_tensor.convert %t2 : tensor<?x?xf64> to tensor<?x?xf64, #DCSR>
%4 = sparse_tensor.convert %t2 : tensor<?x?xf64> to tensor<?x?xf64, #DCSC> %4 = sparse_tensor.convert %t2 : tensor<?x?xf64> to tensor<?x?xf64, #DCSC>
// Two cross conversions. // Two cross conversions.
%5 = sparse_tensor.convert %3 : tensor<?x?xf64, #DCSR> to tensor<?x?xf64, #DCSC> %5 = sparse_tensor.convert %3 : tensor<?x?xf64, #DCSR> to tensor<?x?xf64, #DCSC>
%6 = sparse_tensor.convert %4 : tensor<?x?xf64, #DCSC> to tensor<?x?xf64, #DCSR> %6 = sparse_tensor.convert %4 : tensor<?x?xf64, #DCSC> to tensor<?x?xf64, #DCSR>
// //
// All proper row-/column-wise? // All proper row-/column-wise?
// //
// CHECK: ( 1, 2, 3, 4, 5, 6, 7, -1 ) // CHECK: ( 1, 2, 3, 4, 5, 6, 7, 0 )
aartbikUnsubmitted
Not Done
ReplyInline Actions

alternatively, we could simply print one less, and check the noe field for added safety

aartbik: alternatively, we could simply print one less, and check the noe field for added safety
// CHECK: ( 1, 4, 6, 2, 5, 3, 7, -1 ) // CHECK: ( 1, 4, 6, 2, 5, 3, 7, 0 )
// CHECK: ( 1, 2, 3, 4, 5, 6, 7, -1 ) // CHECK: ( 1, 2, 3, 4, 5, 6, 7, 0 )
// CHECK: ( 1, 4, 6, 2, 5, 3, 7, -1 ) // CHECK: ( 1, 4, 6, 2, 5, 3, 7, 0 )
// CHECK: ( 1, 4, 6, 2, 5, 3, 7, -1 ) // CHECK: ( 1, 4, 6, 2, 5, 3, 7, 0 )
// CHECK: ( 1, 2, 3, 4, 5, 6, 7, -1 ) // CHECK: ( 1, 2, 3, 4, 5, 6, 7, 0 )
// //
%m1 = sparse_tensor.values %1 : tensor<?x?xf64, #DCSR> to memref<?xf64> %m1 = sparse_tensor.values %1 : tensor<?x?xf64, #DCSR> to memref<?xf64>
%m2 = sparse_tensor.values %2 : tensor<?x?xf64, #DCSC> to memref<?xf64> %m2 = sparse_tensor.values %2 : tensor<?x?xf64, #DCSC> to memref<?xf64>
%m3 = sparse_tensor.values %3 : tensor<?x?xf64, #DCSR> to memref<?xf64> %m3 = sparse_tensor.values %3 : tensor<?x?xf64, #DCSR> to memref<?xf64>
%m4 = sparse_tensor.values %4 : tensor<?x?xf64, #DCSC> to memref<?xf64> %m4 = sparse_tensor.values %4 : tensor<?x?xf64, #DCSC> to memref<?xf64>
%m5 = sparse_tensor.values %5 : tensor<?x?xf64, #DCSC> to memref<?xf64> %m5 = sparse_tensor.values %5 : tensor<?x?xf64, #DCSC> to memref<?xf64>
%m6 = sparse_tensor.values %6 : tensor<?x?xf64, #DCSR> to memref<?xf64> %m6 = sparse_tensor.values %6 : tensor<?x?xf64, #DCSR> to memref<?xf64>
call @dump(%m1) : (memref<?xf64>) -> () call @dump(%m1) : (memref<?xf64>) -> ()
Show All 17 Lines

mlir/test/Integration/Dialect/SparseTensor/CPU/sparse_conversion_ptr.mlir

// RUN: mlir-opt %s --sparse-compiler | \ // RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=true | \
// RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s
// RUN: mlir-opt %s --sparse-compiler="enable-runtime-library=false enable-buffer-initialization=true" | \
// RUN: mlir-cpu-runner \ // RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \ // RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \ // RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s // RUN: FileCheck %s
#DCSR = #sparse_tensor.encoding<{ #DCSR = #sparse_tensor.encoding<{
dimLevelType = [ "compressed", "compressed" ], dimLevelType = [ "compressed", "compressed" ],
pointerBitWidth = 8, pointerBitWidth = 8,
Show All 22 Lines
module { module {
// //
// Helper method to print values and indices arrays. The transfer actually // Helper method to print values and indices arrays. The transfer actually
// reads more than required to verify size of buffer as well. // reads more than required to verify size of buffer as well.
// //
func.func @dumpf64(%arg0: memref<?xf64>) { func.func @dumpf64(%arg0: memref<?xf64>) {
%c = arith.constant 0 : index %c = arith.constant 0 : index
%d = arith.constant -1.0 : f64 %d = arith.constant 0.0 : f64
%0 = vector.transfer_read %arg0[%c], %d: memref<?xf64>, vector<8xf64> %0 = vector.transfer_read %arg0[%c], %d: memref<?xf64>, vector<8xf64>
vector.print %0 : vector<8xf64> vector.print %0 : vector<8xf64>
return return
} }
func.func @dumpi08(%arg0: memref<?xi8>) { func.func @dumpi08(%arg0: memref<?xi8>) {
%c = arith.constant 0 : index %c = arith.constant 0 : index
%d = arith.constant -1 : i8 %d = arith.constant 0 : i8
%0 = vector.transfer_read %arg0[%c], %d: memref<?xi8>, vector<8xi8> %0 = vector.transfer_read %arg0[%c], %d: memref<?xi8>, vector<8xi8>
vector.print %0 : vector<8xi8> vector.print %0 : vector<8xi8>
return return
} }
func.func @dumpi32(%arg0: memref<?xi32>) { func.func @dumpi32(%arg0: memref<?xi32>) {
%c = arith.constant 0 : index %c = arith.constant 0 : index
%d = arith.constant -1 : i32 %d = arith.constant 0 : i32
%0 = vector.transfer_read %arg0[%c], %d: memref<?xi32>, vector<8xi32> %0 = vector.transfer_read %arg0[%c], %d: memref<?xi32>, vector<8xi32>
vector.print %0 : vector<8xi32> vector.print %0 : vector<8xi32>
return return
} }
func.func @dumpi64(%arg0: memref<?xi64>) { func.func @dumpi64(%arg0: memref<?xi64>) {
%c = arith.constant 0 : index %c = arith.constant 0 : index
%d = arith.constant -1 : i64 %d = arith.constant 0 : i64
%0 = vector.transfer_read %arg0[%c], %d: memref<?xi64>, vector<8xi64> %0 = vector.transfer_read %arg0[%c], %d: memref<?xi64>, vector<8xi64>
vector.print %0 : vector<8xi64> vector.print %0 : vector<8xi64>
return return
} }
func.func @entry() { func.func @entry() {
%c1 = arith.constant 1 : index %c1 = arith.constant 1 : index
%t1 = arith.constant sparse< %t1 = arith.constant sparse<
Show All 9 Lines func.func @entry() {
// Sparse to sparse. // Sparse to sparse.
%4 = sparse_tensor.convert %1 : tensor<32x64xf64, #DCSR> to tensor<32x64xf64, #DCSC> %4 = sparse_tensor.convert %1 : tensor<32x64xf64, #DCSR> to tensor<32x64xf64, #DCSC>
%5 = sparse_tensor.convert %2 : tensor<32x64xf64, #DCSC> to tensor<32x64xf64, #DCSR> %5 = sparse_tensor.convert %2 : tensor<32x64xf64, #DCSC> to tensor<32x64xf64, #DCSR>
%6 = sparse_tensor.convert %3 : tensor<32x64xf64, #CSC> to tensor<32x64xf64, #DCSR> %6 = sparse_tensor.convert %3 : tensor<32x64xf64, #CSC> to tensor<32x64xf64, #DCSR>
// //
// All proper row-/column-wise? // All proper row-/column-wise?
// //
// CHECK: ( 1, 2, 3, 4, 5, 6, 7, -1 ) // CHECK: ( 1, 2, 3, 4, 5, 6, 7, 0 )
// CHECK-NEXT: ( 1, 4, 6, 2, 5, 3, 7, -1 ) // CHECK-NEXT: ( 1, 4, 6, 2, 5, 3, 7, 0 )
// CHECK-NEXT: ( 1, 4, 6, 2, 5, 3, 7, -1 ) // CHECK-NEXT: ( 1, 4, 6, 2, 5, 3, 7, 0 )
// CHECK-NEXT: ( 1, 4, 6, 2, 5, 3, 7, -1 ) // CHECK-NEXT: ( 1, 4, 6, 2, 5, 3, 7, 0 )
// CHECK-NEXT: ( 1, 2, 3, 4, 5, 6, 7, -1 ) // CHECK-NEXT: ( 1, 2, 3, 4, 5, 6, 7, 0 )
// CHECK-NEXT: ( 1, 2, 3, 4, 5, 6, 7, -1 ) // CHECK-NEXT: ( 1, 2, 3, 4, 5, 6, 7, 0 )
// //
%m1 = sparse_tensor.values %1 : tensor<32x64xf64, #DCSR> to memref<?xf64> %m1 = sparse_tensor.values %1 : tensor<32x64xf64, #DCSR> to memref<?xf64>
%m2 = sparse_tensor.values %2 : tensor<32x64xf64, #DCSC> to memref<?xf64> %m2 = sparse_tensor.values %2 : tensor<32x64xf64, #DCSC> to memref<?xf64>
%m3 = sparse_tensor.values %3 : tensor<32x64xf64, #CSC> to memref<?xf64> %m3 = sparse_tensor.values %3 : tensor<32x64xf64, #CSC> to memref<?xf64>
%m4 = sparse_tensor.values %4 : tensor<32x64xf64, #DCSC> to memref<?xf64> %m4 = sparse_tensor.values %4 : tensor<32x64xf64, #DCSC> to memref<?xf64>
%m5 = sparse_tensor.values %5 : tensor<32x64xf64, #DCSR> to memref<?xf64> %m5 = sparse_tensor.values %5 : tensor<32x64xf64, #DCSR> to memref<?xf64>
%m6 = sparse_tensor.values %6 : tensor<32x64xf64, #DCSR> to memref<?xf64> %m6 = sparse_tensor.values %6 : tensor<32x64xf64, #DCSR> to memref<?xf64>
call @dumpf64(%m1) : (memref<?xf64>) -> () call @dumpf64(%m1) : (memref<?xf64>) -> ()
call @dumpf64(%m2) : (memref<?xf64>) -> () call @dumpf64(%m2) : (memref<?xf64>) -> ()
call @dumpf64(%m3) : (memref<?xf64>) -> () call @dumpf64(%m3) : (memref<?xf64>) -> ()
call @dumpf64(%m4) : (memref<?xf64>) -> () call @dumpf64(%m4) : (memref<?xf64>) -> ()
call @dumpf64(%m5) : (memref<?xf64>) -> () call @dumpf64(%m5) : (memref<?xf64>) -> ()
call @dumpf64(%m6) : (memref<?xf64>) -> () call @dumpf64(%m6) : (memref<?xf64>) -> ()
// //
// Sanity check on indices. // Sanity check on indices.
// //
// CHECK-NEXT: ( 0, 1, 63, 0, 1, 0, 63, -1 ) // CHECK-NEXT: ( 0, 1, 63, 0, 1, 0, 63, 0 )
// CHECK-NEXT: ( 0, 1, 31, 0, 1, 0, 31, -1 ) // CHECK-NEXT: ( 0, 1, 31, 0, 1, 0, 31, 0 )
// CHECK-NEXT: ( 0, 1, 31, 0, 1, 0, 31, -1 ) // CHECK-NEXT: ( 0, 1, 31, 0, 1, 0, 31, 0 )
// CHECK-NEXT: ( 0, 1, 31, 0, 1, 0, 31, -1 ) // CHECK-NEXT: ( 0, 1, 31, 0, 1, 0, 31, 0 )
// CHECK-NEXT: ( 0, 1, 63, 0, 1, 0, 63, -1 ) // CHECK-NEXT: ( 0, 1, 63, 0, 1, 0, 63, 0 )
// CHECK-NEXT: ( 0, 1, 63, 0, 1, 0, 63, -1 ) // CHECK-NEXT: ( 0, 1, 63, 0, 1, 0, 63, 0 )
// //
%i1 = sparse_tensor.indices %1 { dimension = 1 : index } : tensor<32x64xf64, #DCSR> to memref<?xi8> %i1 = sparse_tensor.indices %1 { dimension = 1 : index } : tensor<32x64xf64, #DCSR> to memref<?xi8>
%i2 = sparse_tensor.indices %2 { dimension = 1 : index } : tensor<32x64xf64, #DCSC> to memref<?xi64> %i2 = sparse_tensor.indices %2 { dimension = 1 : index } : tensor<32x64xf64, #DCSC> to memref<?xi64>
%i3 = sparse_tensor.indices %3 { dimension = 1 : index } : tensor<32x64xf64, #CSC> to memref<?xi32> %i3 = sparse_tensor.indices %3 { dimension = 1 : index } : tensor<32x64xf64, #CSC> to memref<?xi32>
%i4 = sparse_tensor.indices %4 { dimension = 1 : index } : tensor<32x64xf64, #DCSC> to memref<?xi64> %i4 = sparse_tensor.indices %4 { dimension = 1 : index } : tensor<32x64xf64, #DCSC> to memref<?xi64>
%i5 = sparse_tensor.indices %5 { dimension = 1 : index } : tensor<32x64xf64, #DCSR> to memref<?xi8> %i5 = sparse_tensor.indices %5 { dimension = 1 : index } : tensor<32x64xf64, #DCSR> to memref<?xi8>
%i6 = sparse_tensor.indices %6 { dimension = 1 : index } : tensor<32x64xf64, #DCSR> to memref<?xi8> %i6 = sparse_tensor.indices %6 { dimension = 1 : index } : tensor<32x64xf64, #DCSR> to memref<?xi8>
call @dumpi08(%i1) : (memref<?xi8>) -> () call @dumpi08(%i1) : (memref<?xi8>) -> ()
Show All 17 Lines

mlir/test/Integration/Dialect/SparseTensor/CPU/sparse_matrix_ops.mlir

// RUN: mlir-opt %s --sparse-compiler | \ // RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=true | \
// RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s
// RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=false | \
// RUN: mlir-cpu-runner \ // RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \ // RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \ // RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s // RUN: FileCheck %s
#DCSR = #sparse_tensor.encoding<{dimLevelType = ["compressed", "compressed"]}> #DCSR = #sparse_tensor.encoding<{dimLevelType = ["compressed", "compressed"]}>
// //
▲ Show 20 Lines • Show All 160 Lines • Show Last 20 Lines

mlir/test/Integration/Dialect/SparseTensor/CPU/sparse_out_mult_elt.mlir

// RUN: mlir-opt %s --sparse-compiler | \ // RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=true | \
// RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s
// RUN: mlir-opt %s --sparse-compiler="enable-runtime-library=false enable-buffer-initialization=true"| \
// RUN: mlir-cpu-runner \ // RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \ // RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \ // RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s // RUN: FileCheck %s
#DCSR = #sparse_tensor.encoding<{ #DCSR = #sparse_tensor.encoding<{
dimLevelType = [ "compressed", "compressed" ] dimLevelType = [ "compressed", "compressed" ]
}> }>
Show All 21 Lines %0 = linalg.generic #trait_mult_elt
linalg.yield %1 : f32 linalg.yield %1 : f32
} -> tensor<32x16xf32, #DCSR> } -> tensor<32x16xf32, #DCSR>
return %0 : tensor<32x16xf32, #DCSR> return %0 : tensor<32x16xf32, #DCSR>
} }
// Driver method to call and verify kernel. // Driver method to call and verify kernel.
func.func @entry() { func.func @entry() {
%c0 = arith.constant 0 : index %c0 = arith.constant 0 : index
%f1 = arith.constant -1.0 : f32 %f0 = arith.constant 0.0 : f32
// Setup very sparse matrices. // Setup very sparse matrices.
%ta = arith.constant sparse< %ta = arith.constant sparse<
[ [2,2], [15,15], [31,0], [31,14] ], [ 2.0, 3.0, -2.0, 4.0 ] [ [2,2], [15,15], [31,0], [31,14] ], [ 2.0, 3.0, -2.0, 4.0 ]
> : tensor<32x16xf32> > : tensor<32x16xf32>
%tb = arith.constant sparse< %tb = arith.constant sparse<
[ [1,1], [2,0], [2,2], [2,15], [31,0], [31,15] ], [ 5.0, 6.0, 7.0, 8.0, -10.0, 9.0 ] [ [1,1], [2,0], [2,2], [2,15], [31,0], [31,15] ], [ 5.0, 6.0, 7.0, 8.0, -10.0, 9.0 ]
> : tensor<32x16xf32> > : tensor<32x16xf32>
%sta = sparse_tensor.convert %ta %sta = sparse_tensor.convert %ta
: tensor<32x16xf32> to tensor<32x16xf32, #DCSR> : tensor<32x16xf32> to tensor<32x16xf32, #DCSR>
%stb = sparse_tensor.convert %tb %stb = sparse_tensor.convert %tb
: tensor<32x16xf32> to tensor<32x16xf32, #DCSR> : tensor<32x16xf32> to tensor<32x16xf32, #DCSR>
// Call kernel. // Call kernel.
%0 = call @sparse_mult_elt(%sta, %stb) %0 = call @sparse_mult_elt(%sta, %stb)
: (tensor<32x16xf32, #DCSR>, : (tensor<32x16xf32, #DCSR>,
tensor<32x16xf32, #DCSR>) -> tensor<32x16xf32, #DCSR> tensor<32x16xf32, #DCSR>) -> tensor<32x16xf32, #DCSR>
// //
// Verify results. Only two entries stored in result! // Verify results. Only two entries stored in result!
// //
// CHECK: ( 14, 20, -1, -1 ) // CHECK: ( 14, 20, 0, 0 )
// //
%val = sparse_tensor.values %0 : tensor<32x16xf32, #DCSR> to memref<?xf32> %val = sparse_tensor.values %0 : tensor<32x16xf32, #DCSR> to memref<?xf32>
%vv = vector.transfer_read %val[%c0], %f1: memref<?xf32>, vector<4xf32> %vv = vector.transfer_read %val[%c0], %f0: memref<?xf32>, vector<4xf32>
vector.print %vv : vector<4xf32> vector.print %vv : vector<4xf32>
// Release the resources. // Release the resources.
bufferization.dealloc_tensor %sta : tensor<32x16xf32, #DCSR> bufferization.dealloc_tensor %sta : tensor<32x16xf32, #DCSR>
bufferization.dealloc_tensor %stb : tensor<32x16xf32, #DCSR> bufferization.dealloc_tensor %stb : tensor<32x16xf32, #DCSR>
bufferization.dealloc_tensor %0 : tensor<32x16xf32, #DCSR> bufferization.dealloc_tensor %0 : tensor<32x16xf32, #DCSR>
return return
} }
} }

mlir/test/Integration/Dialect/SparseTensor/CPU/sparse_out_reduction.mlir

// RUN: mlir-opt %s --sparse-compiler | \ // RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=true | \
// RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s
// RUN: mlir-opt %s --sparse-compiler="enable-runtime-library=false enable-buffer-initialization=true"| \
// RUN: mlir-cpu-runner \ // RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \ // RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \ // RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s // RUN: FileCheck %s
#SparseMatrix = #sparse_tensor.encoding<{ #SparseMatrix = #sparse_tensor.encoding<{
dimLevelType = [ "compressed", "compressed" ] dimLevelType = [ "compressed", "compressed" ]
}> }>
Show All 31 Lines %0 = linalg.generic #redsum
linalg.yield %1 : i32 linalg.yield %1 : i32
} -> tensor<?x?xi32, #SparseMatrix> } -> tensor<?x?xi32, #SparseMatrix>
return %0 : tensor<?x?xi32, #SparseMatrix> return %0 : tensor<?x?xi32, #SparseMatrix>
} }
// Driver method to call and verify tensor kernel. // Driver method to call and verify tensor kernel.
func.func @entry() { func.func @entry() {
%c0 = arith.constant 0 : index %c0 = arith.constant 0 : index
%i0 = arith.constant -1 : i32 %i0 = arith.constant 0 : i32
// Setup very sparse 3-d tensors. // Setup very sparse 3-d tensors.
%t1 = arith.constant sparse< %t1 = arith.constant sparse<
[ [1,1,3], [2,0,0], [2,2,1], [2,2,2], [2,2,3] ], [ 1, 2, 3, 4, 5 ] [ [1,1,3], [2,0,0], [2,2,1], [2,2,2], [2,2,3] ], [ 1, 2, 3, 4, 5 ]
> : tensor<3x3x4xi32> > : tensor<3x3x4xi32>
%t2 = arith.constant sparse< %t2 = arith.constant sparse<
[ [1,0,0], [1,1,3], [2,2,1], [2,2,3] ], [ 6, 7, 8, 9 ] [ [1,0,0], [1,1,3], [2,2,1], [2,2,3] ], [ 6, 7, 8, 9 ]
> : tensor<3x3x4xi32> > : tensor<3x3x4xi32>
%st1 = sparse_tensor.convert %t1 %st1 = sparse_tensor.convert %t1
: tensor<3x3x4xi32> to tensor<?x?x?xi32, #SparseTensor> : tensor<3x3x4xi32> to tensor<?x?x?xi32, #SparseTensor>
%st2 = sparse_tensor.convert %t2 %st2 = sparse_tensor.convert %t2
: tensor<3x3x4xi32> to tensor<?x?x?xi32, #SparseTensor> : tensor<3x3x4xi32> to tensor<?x?x?xi32, #SparseTensor>
// Call kernel. // Call kernel.
%0 = call @redsum(%st1, %st2) %0 = call @redsum(%st1, %st2)
: (tensor<?x?x?xi32, #SparseTensor>, : (tensor<?x?x?xi32, #SparseTensor>,
tensor<?x?x?xi32, #SparseTensor>) -> tensor<?x?xi32, #SparseMatrix> tensor<?x?x?xi32, #SparseTensor>) -> tensor<?x?xi32, #SparseMatrix>
// //
// Verify results. Only two entries stored in result. Correct structure. // Verify results. Only two entries stored in result. Correct structure.
// //
// CHECK: ( 7, 69, -1, -1 ) // CHECK: ( 7, 69, 0, 0 )
// CHECK-NEXT: ( ( 0, 0, 0 ), ( 0, 7, 0 ), ( 0, 0, 69 ) ) // CHECK-NEXT: ( ( 0, 0, 0 ), ( 0, 7, 0 ), ( 0, 0, 69 ) )
// //
%val = sparse_tensor.values %0 %val = sparse_tensor.values %0
: tensor<?x?xi32, #SparseMatrix> to memref<?xi32> : tensor<?x?xi32, #SparseMatrix> to memref<?xi32>
%vv = vector.transfer_read %val[%c0], %i0: memref<?xi32>, vector<4xi32> %vv = vector.transfer_read %val[%c0], %i0: memref<?xi32>, vector<4xi32>
vector.print %vv : vector<4xi32> vector.print %vv : vector<4xi32>
%dm = sparse_tensor.convert %0 %dm = sparse_tensor.convert %0
: tensor<?x?xi32, #SparseMatrix> to tensor<?x?xi32> : tensor<?x?xi32, #SparseMatrix> to tensor<?x?xi32>
Show All 10 Lines

mlir/test/Integration/Dialect/SparseTensor/CPU/sparse_storage.mlir

// RUN: mlir-opt %s --sparse-compiler | \ // RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=true | \
// RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s
// RUN: mlir-opt %s --sparse-compiler=enable-runtime-library=false | \
// RUN: mlir-cpu-runner \ // RUN: mlir-cpu-runner \
// RUN: -e entry -entry-point-result=void \ // RUN: -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \ // RUN: -shared-libs=%mlir_lib_dir/libmlir_c_runner_utils%shlibext | \
// RUN: FileCheck %s // RUN: FileCheck %s
// //
// Several common sparse storage schemes. // Several common sparse storage schemes.
// //
▲ Show 20 Lines • Show All 250 Lines • Show Last 20 Lines