Debian Codegraph @

flang/test/Fir/CUDA/cuda-device-context.mlir

raw
1 // RUN: fir-opt --simplify-intrinsics %s | FileCheck %s
2
3 func.func @_QPsum_in_device(%arg0: !fir.ref<!fir.array<?xi32>> {cuf.data_attr = #cuf.cuda<device>, fir.bindc_name = "a"}, %arg1: i32 {fir.bindc_name = "n"}) attributes {cuf.proc_attr = #cuf.cuda_proc<global>} {
4 %c5_i32 = arith.constant 5 : i32
5 %c1 = arith.constant 1 : index
6 %c0 = arith.constant 0 : index
7 %c-1 = arith.constant -1 : index
8 %0 = fir.dummy_scope : !fir.dscope
9 %1 = fir.shape %c-1 : (index) -> !fir.shape<1>
10 %2 = fir.declare %arg0(%1) dummy_scope %0 {data_attr = #cuf.cuda<device>, uniq_name = "_QFsum_in_deviceEa"} : (!fir.ref<!fir.array<?xi32>>, !fir.shape<1>, !fir.dscope) -> !fir.ref<!fir.array<?xi32>>
11 %3 = fir.embox %2(%1) : (!fir.ref<!fir.array<?xi32>>, !fir.shape<1>) -> !fir.box<!fir.array<?xi32>>
12 %4 = fir.alloca i32
13 fir.store %arg1 to %4 : !fir.ref<i32>
14 %5 = fir.declare %4 dummy_scope %0 {fortran_attrs = #fir.var_attrs<value>, uniq_name = "_QFsum_in_deviceEn"} : (!fir.ref<i32>, !fir.dscope) -> !fir.ref<i32>
15 %12 = fir.alloca i32 {bindc_name = "i", uniq_name = "_QFsum_in_deviceEi"}
16 %13 = fir.declare %12 {uniq_name = "_QFsum_in_deviceEi"} : (!fir.ref<i32>) -> !fir.ref<i32>
17 %14 = fir.address_of(@_QM__fortran_builtinsE__builtin_threadidx) : !fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>
18 %18 = fir.load %5 : !fir.ref<i32>
19 %19 = fir.convert %18 : (i32) -> index
20 %20 = arith.cmpi sgt, %19, %c0 : index
21 %21 = arith.select %20, %19, %c0 : index
22 %22 = fir.alloca !fir.array<?xi32>, %21 {bindc_name = "auto", uniq_name = "_QFsum_in_deviceEauto"}
23 %23 = fir.shape %21 : (index) -> !fir.shape<1>
24 %24 = fir.declare %22(%23) {uniq_name = "_QFsum_in_deviceEauto"} : (!fir.ref<!fir.array<?xi32>>, !fir.shape<1>) -> !fir.ref<!fir.array<?xi32>>
25 %25 = fir.embox %24(%23) : (!fir.ref<!fir.array<?xi32>>, !fir.shape<1>) -> !fir.box<!fir.array<?xi32>>
26 %26 = fir.undefined index
27 %27 = fir.slice %c1, %19, %c1 : (index, index, index) -> !fir.slice<1>
28 %28 = fir.embox %24(%23) [%27] : (!fir.ref<!fir.array<?xi32>>, !fir.shape<1>, !fir.slice<1>) -> !fir.box<!fir.array<?xi32>>
29 %29 = fir.absent !fir.box<i1>
30 %30 = fir.address_of(@_QQclX91d13f6e74caa2f03965d7a7c6a8fdd5) : !fir.ref<!fir.char<1,50>>
31 %31 = fir.convert %28 : (!fir.box<!fir.array<?xi32>>) -> !fir.box<none>
32 %32 = fir.convert %30 : (!fir.ref<!fir.char<1,50>>) -> !fir.ref<i8>
33 %33 = fir.convert %c0 : (index) -> i32
34 %34 = fir.convert %29 : (!fir.box<i1>) -> !fir.box<none>
35 %35 = fir.call @_FortranASumInteger4(%31, %32, %c5_i32, %33, %34) fastmath<contract> : (!fir.box<none>, !fir.ref<i8>, i32, i32, !fir.box<none>) -> i32
36 %36 = fir.load %13 : !fir.ref<i32>
37 %37 = fir.convert %36 : (i32) -> i64
38 %38 = fir.array_coor %2(%1) %37 : (!fir.ref<!fir.array<?xi32>>, !fir.shape<1>, i64) -> !fir.ref<i32>
39 fir.store %35 to %38 : !fir.ref<i32>
40 return
41 }
42
43 // Check that intrinsic simplification is disabled in CUDA Fortran context. The simplified intrinsic is
44 // created in the module op but the device func will be migrated into a gpu module op resulting in a
45 // missing symbol error.
46 // The simplified intrinsic could also be migrated to the gpu module but the choice has not be made
47 // at this point.
48 // CHECK-LABEL: func.func @_QPsum_in_device
49 // CHECK-NOT: fir.call @_FortranASumInteger4x1_contract_simplified