Files
TheRock/compiler/amd-llvm/flang/test/Lower/CUDA/cuda-kernel-calls.cuf
T
derek.holloway 7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
Flatten subprojects into a single base monorepo
2026-09-02 16:34:47 -07:00

68 lines
4.9 KiB
Plaintext

! RUN: bbc -emit-hlfir -fcuda %s -o - | FileCheck %s
! RUN: bbc -emit-hlfir -fcuda %s -o - | fir-opt | FileCheck %s
! Test lowering of CUDA procedure calls.
module test_call
use, intrinsic :: __fortran_builtins, only: __builtin_dim3
contains
attributes(global) subroutine dev_kernel0()
end
attributes(global) subroutine dev_kernel1(a)
real :: a
end
subroutine host()
real, device :: a
integer(8) :: stream
integer(4) :: nbytes
! CHECK-LABEL: func.func @_QMtest_callPhost()
! CHECK: %[[A:.*]]:2 = hlfir.declare %{{.*}} {data_attr = #cuf.cuda<device>, uniq_name = "_QMtest_callFhostEa"} : (!fir.ref<f32>) -> (!fir.ref<f32>, !fir.ref<f32>)
call dev_kernel0<<<10, 20>>>()
! CHECK: cuf.kernel_launch @_QMtest_callPdev_kernel0<<<%c10{{.*}}, %c1{{.*}}, %c1{{.*}}, %c20{{.*}}, %c1{{.*}}, %c1{{.*}}>>>()
call dev_kernel0<<< __builtin_dim3(1,1,4), __builtin_dim3(32,1,1) >>>
! CHECK: %[[ADDR_DIM3_GRID:.*]] = fir.address_of(@_QQro._QM__fortran_builtinsT__builtin_dim3.{{.*}}) : !fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>
! CHECK: %[[DIM3_GRID:.*]]:2 = hlfir.declare %[[ADDR_DIM3_GRID]] {fortran_attrs = #fir.var_attrs<parameter>, uniq_name = "_QQro._QM__fortran_builtinsT__builtin_dim3.0"} : (!fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>) -> (!fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>, !fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>)
! CHECK: %[[GRID_X:.*]] = hlfir.designate %[[DIM3_GRID]]#0{"x"} : (!fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>) -> !fir.ref<i32>
! CHECK: %[[GRID_X_LOAD:.*]] = fir.load %[[GRID_X]] : !fir.ref<i32>
! CHECK: %[[GRID_Y:.*]] = hlfir.designate %[[DIM3_GRID]]#0{"y"} : (!fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>) -> !fir.ref<i32>
! CHECK: %[[GRID_Y_LOAD:.*]] = fir.load %[[GRID_Y]] : !fir.ref<i32>
! CHECK: %[[GRID_Z:.*]] = hlfir.designate %[[DIM3_GRID]]#0{"z"} : (!fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>) -> !fir.ref<i32>
! CHECK: %[[GRID_Z_LOAD:.*]] = fir.load %[[GRID_Z]] : !fir.ref<i32>
! CHECK: %[[ADDR_DIM3_BLOCK:.*]] = fir.address_of(@_QQro._QM__fortran_builtinsT__builtin_dim3.{{.*}}) : !fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>
! CHECK: %[[DIM3_BLOCK:.*]]:2 = hlfir.declare %[[ADDR_DIM3_BLOCK]] {fortran_attrs = #fir.var_attrs<parameter>, uniq_name = "_QQro._QM__fortran_builtinsT__builtin_dim3.1"} : (!fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>) -> (!fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>, !fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>)
! CHECK: %[[BLOCK_X:.*]] = hlfir.designate %[[DIM3_BLOCK]]#0{"x"} : (!fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>) -> !fir.ref<i32>
! CHECK: %[[BLOCK_X_LOAD:.*]] = fir.load %[[BLOCK_X]] : !fir.ref<i32>
! CHECK: %[[BLOCK_Y:.*]] = hlfir.designate %[[DIM3_BLOCK]]#0{"y"} : (!fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>) -> !fir.ref<i32>
! CHECK: %[[BLOCK_Y_LOAD:.*]] = fir.load %[[BLOCK_Y]] : !fir.ref<i32>
! CHECK: %[[BLOCK_Z:.*]] = hlfir.designate %[[DIM3_BLOCK]]#0{"z"} : (!fir.ref<!fir.type<_QM__fortran_builtinsT__builtin_dim3{x:i32,y:i32,z:i32}>>) -> !fir.ref<i32>
! CHECK: %[[BLOCK_Z_LOAD:.*]] = fir.load %[[BLOCK_Z]] : !fir.ref<i32>
! CHECK: cuf.kernel_launch @_QMtest_callPdev_kernel0<<<%[[GRID_X_LOAD]], %[[GRID_Y_LOAD]], %[[GRID_Z_LOAD]], %[[BLOCK_X_LOAD]], %[[BLOCK_Y_LOAD]], %[[BLOCK_Z_LOAD]]>>>()
call dev_kernel0<<<10, 20, 2>>>()
! CHECK: cuf.kernel_launch @_QMtest_callPdev_kernel0<<<%c10{{.*}}, %c1{{.*}}, %c1{{.*}}, %c20{{.*}}, %c1{{.*}}, %c1{{.*}}, %c2{{.*}}>>>()
call dev_kernel0<<<10, 20, 2, 0_8>>>()
! CHECK: cuf.kernel_launch @_QMtest_callPdev_kernel0<<<%c10{{.*}}, %c1{{.*}}, %c1{{.*}}, %c20{{.*}}, %c1{{.*}}, %c1{{.*}}, %c2{{.*}}, %{{.*}} : !fir.ref<i64>>>>()
call dev_kernel1<<<1, 32>>>(a)
! CHECK: cuf.kernel_launch @_QMtest_callPdev_kernel1<<<%c1{{.*}}, %c1{{.*}}, %c1{{.*}}, %c32{{.*}}, %c1{{.*}}, %c1{{.*}}>>>(%{{.*}}) : (!fir.ref<f32>)
call dev_kernel1<<<*, 32>>>(a)
! CHECK: cuf.kernel_launch @_QMtest_callPdev_kernel1<<<%c-1{{.*}}, %c1{{.*}}, %c1{{.*}}, %c32{{.*}}, %c1{{.*}}, %c1{{.*}}>>>(%{{.*}})
call dev_kernel1<<<*,32,0,stream>>>(a)
! CHECK: cuf.kernel_launch @_QMtest_callPdev_kernel1<<<%c-1{{.*}}, %c1{{.*}}, %c1{{.*}}, %c32{{.*}}, %c1{{.*}}, %c1{{.*}}, %c0{{.*}}, %{{.*}} : !fir.ref<i64>>>>(%{{.*}}) : (!fir.ref<f32>)
call dev_kernel1<<<*, 32, 0.8 * nbytes>>>(a)
! CHECK: %[[MUL:.*]] = arith.mulf %{{.*}}, %{{.*}} fastmath<contract> : f32
! CHECK: %[[BYTES:.*]] = fir.convert %[[MUL]] : (f32) -> i32
! CHECK: cuf.kernel_launch @_QMtest_callPdev_kernel1<<<%{{.*}}, %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}}, %[[BYTES]]>>>(%{{.*}}) : (!fir.ref<f32>)
end
end