Files
clang-p2996/mlir/test/Integration/Dialect/Vector/CPU/test-transfer-read.mlir
Markus Böck 9048ea28da Reland "[mlir] Make the vast majority of intgration and runner tests work on Windows"
This reverts commit 5561e17411

The logic was moved from cmake into lit fixing the issue that lead to the revert and potentially others with multi-config cmake generators

Differential Revision: https://reviews.llvm.org/D143925
2023-02-15 19:14:43 +01:00

96 lines
3.8 KiB
MLIR

// RUN: mlir-opt %s -pass-pipeline="builtin.module(func.func(convert-vector-to-scf,lower-affine,convert-scf-to-cf),convert-vector-to-llvm,finalize-memref-to-llvm,convert-func-to-llvm,reconcile-unrealized-casts)" | \
// RUN: mlir-cpu-runner -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_c_runner_utils | \
// RUN: FileCheck %s
// RUN: mlir-opt %s -pass-pipeline="builtin.module(func.func(convert-vector-to-scf{full-unroll=true},lower-affine,convert-scf-to-cf),convert-vector-to-llvm,finalize-memref-to-llvm,convert-func-to-llvm,reconcile-unrealized-casts)" | \
// RUN: mlir-cpu-runner -e entry -entry-point-result=void \
// RUN: -shared-libs=%mlir_c_runner_utils | \
// RUN: FileCheck %s
func.func @transfer_read_1d(%A : memref<?xf32>, %base: index) {
%fm42 = arith.constant -42.0: f32
%f = vector.transfer_read %A[%base], %fm42
{permutation_map = affine_map<(d0) -> (d0)>} :
memref<?xf32>, vector<13xf32>
vector.print %f: vector<13xf32>
return
}
func.func @transfer_read_mask_1d(%A : memref<?xf32>, %base: index) {
%fm42 = arith.constant -42.0: f32
%m = arith.constant dense<[0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0]> : vector<13xi1>
%f = vector.transfer_read %A[%base], %fm42, %m : memref<?xf32>, vector<13xf32>
vector.print %f: vector<13xf32>
return
}
func.func @transfer_read_inbounds_4(%A : memref<?xf32>, %base: index) {
%fm42 = arith.constant -42.0: f32
%f = vector.transfer_read %A[%base], %fm42
{permutation_map = affine_map<(d0) -> (d0)>, in_bounds = [true]} :
memref<?xf32>, vector<4xf32>
vector.print %f: vector<4xf32>
return
}
func.func @transfer_read_mask_inbounds_4(%A : memref<?xf32>, %base: index) {
%fm42 = arith.constant -42.0: f32
%m = arith.constant dense<[0, 1, 0, 1]> : vector<4xi1>
%f = vector.transfer_read %A[%base], %fm42, %m {in_bounds = [true]}
: memref<?xf32>, vector<4xf32>
vector.print %f: vector<4xf32>
return
}
func.func @transfer_write_1d(%A : memref<?xf32>, %base: index) {
%f0 = arith.constant 0.0 : f32
%vf0 = vector.splat %f0 : vector<4xf32>
vector.transfer_write %vf0, %A[%base]
{permutation_map = affine_map<(d0) -> (d0)>} :
vector<4xf32>, memref<?xf32>
return
}
func.func @entry() {
%c0 = arith.constant 0: index
%c1 = arith.constant 1: index
%c2 = arith.constant 2: index
%c3 = arith.constant 3: index
%c4 = arith.constant 4: index
%c5 = arith.constant 5: index
%A = memref.alloc(%c5) : memref<?xf32>
scf.for %i = %c0 to %c5 step %c1 {
%i32 = arith.index_cast %i : index to i32
%fi = arith.sitofp %i32 : i32 to f32
memref.store %fi, %A[%i] : memref<?xf32>
}
// On input, memory contains [[ 0, 1, 2, 3, 4, xxx garbage xxx ]]
// Read shifted by 2 and pad with -42:
// ( 2, 3, 4, -42, ..., -42)
call @transfer_read_1d(%A, %c2) : (memref<?xf32>, index) -> ()
// Read with mask and out-of-bounds access.
call @transfer_read_mask_1d(%A, %c2) : (memref<?xf32>, index) -> ()
// Write into memory shifted by 3
// memory contains [[ 0, 1, 2, 0, 0, xxx garbage xxx ]]
call @transfer_write_1d(%A, %c3) : (memref<?xf32>, index) -> ()
// Read shifted by 0 and pad with -42:
// ( 0, 1, 2, 0, 0, -42, ..., -42)
call @transfer_read_1d(%A, %c0) : (memref<?xf32>, index) -> ()
// Read in-bounds 4 @ 1, guaranteed to not overflow.
// Exercises proper alignment.
call @transfer_read_inbounds_4(%A, %c1) : (memref<?xf32>, index) -> ()
// Read in-bounds with mask.
call @transfer_read_mask_inbounds_4(%A, %c1) : (memref<?xf32>, index) -> ()
memref.dealloc %A : memref<?xf32>
return
}
// CHECK: ( 2, 3, 4, -42, -42, -42, -42, -42, -42, -42, -42, -42, -42 )
// CHECK: ( -42, -42, 4, -42, -42, -42, -42, -42, -42, -42, -42, -42, -42 )
// CHECK: ( 0, 1, 2, 0, 0, -42, -42, -42, -42, -42, -42, -42, -42 )
// CHECK: ( 1, 2, 0, 0 )
// CHECK: ( -42, 2, -42, 0 )