1 // RUN: mlir-opt -convert-affine-for-to-gpu="gpu-block-dims=0 gpu-thread-dims=1" %s | FileCheck --check-prefix=CHECK-THREADS %s
2 // RUN: mlir-opt -convert-affine-for-to-gpu="gpu-block-dims=1 gpu-thread-dims=0" %s | FileCheck --check-prefix=CHECK-BLOCKS %s
4 // CHECK-THREADS-LABEL: @one_d_loop
5 // CHECK-BLOCKS-LABEL: @one_d_loop
6 func @one_d_loop(%A : memref<?xf32>, %B : memref<?xf32>) {
7 // Bounds of the loop, its range and step.
8 // CHECK-THREADS-NEXT: %{{.*}} = arith.constant 0 : index
9 // CHECK-THREADS-NEXT: %{{.*}} = arith.constant 42 : index
10 // CHECK-THREADS-NEXT: %[[BOUND:.*]] = arith.subi %{{.*}}, %{{.*}} : index
11 // CHECK-THREADS-NEXT: %{{.*}} = arith.constant 1 : index
12 // CHECK-THREADS-NEXT: %[[ONE:.*]] = arith.constant 1 : index
14 // CHECK-BLOCKS-NEXT: %{{.*}} = arith.constant 0 : index
15 // CHECK-BLOCKS-NEXT: %{{.*}} = arith.constant 42 : index
16 // CHECK-BLOCKS-NEXT: %[[BOUND:.*]] = arith.subi %{{.*}}, %{{.*}} : index
17 // CHECK-BLOCKS-NEXT: %{{.*}} = arith.constant 1 : index
18 // CHECK-BLOCKS-NEXT: %[[ONE:.*]] = arith.constant 1 : index
20 // CHECK-THREADS-NEXT: gpu.launch blocks(%[[B0:.*]], %[[B1:.*]], %[[B2:.*]]) in (%{{.*}} = %[[ONE]], %{{.*}} = %[[ONE]], %{{.*}}0 = %[[ONE]]) threads(%[[T0:.*]], %[[T1:.*]], %[[T2:.*]]) in (%{{.*}} = %[[BOUND]], %{{.*}} = %[[ONE]], %{{.*}} = %[[ONE]])
21 // CHECK-BLOCKS-NEXT: gpu.launch blocks(%[[B0:.*]], %[[B1:.*]], %[[B2:.*]]) in (%{{.*}} = %[[BOUND]], %{{.*}} = %[[ONE]], %{{.*}}0 = %[[ONE]]) threads(%[[T0:.*]], %[[T1:.*]], %[[T2:.*]]) in (%{{.*}} = %[[ONE]], %{{.*}} = %[[ONE]], %{{.*}} = %[[ONE]])
22 affine.for %i = 0 to 42 {
23 // CHECK-THREADS-NEXT: %[[INDEX:.*]] = arith.addi %{{.*}}, %[[T0]]
24 // CHECK-THREADS-NEXT: memref.load %{{.*}}[%[[INDEX]]]
25 // CHECK-BLOCKS-NEXT: %[[INDEX:.*]] = arith.addi %{{.*}}, %[[B0]]
26 // CHECK-BLOCKS-NEXT: memref.load %{{.*}}[%[[INDEX]]]
27 %0 = memref.load %A[%i] : memref<?xf32>
28 memref.store %0, %B[%i] : memref<?xf32>
29 // CHECK-THREADS: gpu.terminator
30 // CHECK-BLOCKS: gpu.terminator