Skip to content

[flang][cuda] Update FIROps.td to add $grid_z to CudaKernelLaunch #85318

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Mar 15, 2024
Merged

[flang][cuda] Update FIROps.td to add $grid_z to CudaKernelLaunch #85318

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
a10d6a3
Update FIROps.td to add $grid_z to CudaKernelLaunch
ImanHosseini Mar 14, 2024
4a46909
Update ConvertCall.cpp for 3d grid
ImanHosseini Mar 14, 2024
87eb858
Update cuda-kernel-calls.cuf for 3d grids
ImanHosseini Mar 14, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 3 additions & 2 deletions flang/include/flang/Optimizer/Dialect/FIROps.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2454,6 +2454,7 @@ def fir_CUDAKernelLaunch : fir_Op<"cuda_kernel_launch", [CallOpInterface,
SymbolRefAttr:$callee,
I32:$grid_x,
I32:$grid_y,
I32:$grid_z,
I32:$block_x,
I32:$block_y,
I32:$block_z,
Expand All @@ -2463,8 +2464,8 @@ def fir_CUDAKernelLaunch : fir_Op<"cuda_kernel_launch", [CallOpInterface,
);

let assemblyFormat = [{
$callee `<` `<` `<` $grid_x `,` $grid_y `,` $block_x `,` $block_y `,`
$block_z ( `,` $bytes^ ( `,` $stream^ )? )? `>` `>` `>`
$callee `<` `<` `<` $grid_x `,` $grid_y `,` $grid_z `,`$block_x `,`
$block_y `,` $block_z ( `,` $bytes^ ( `,` $stream^ )? )? `>` `>` `>`
`` `(` ( $args^ `:` type($args) )? `)` attr-dict
}];

Expand Down
4 changes: 2 additions & 2 deletions flang/lib/Lower/ConvertCall.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -466,8 +466,8 @@ std::pair<fir::ExtendedValue, bool> Fortran::lower::genCallOpAndResult(
caller.getCallDescription().chevrons()[3], stmtCtx)));

builder.create<fir::CUDAKernelLaunch>(
loc, funcType.getResults(), funcSymbolAttr, grid_x, grid_y, block_x,
block_y, block_z, bytes, stream, operands);
loc, funcType.getResults(), funcSymbolAttr, grid_x, grid_y, one,
block_x, block_y, block_z, bytes, stream, operands);
callNumResults = 0;
} else if (caller.requireDispatchCall()) {
// Procedure call requiring a dynamic dispatch. Call is created with
Expand Down
10 changes: 5 additions & 5 deletions flang/test/Lower/CUDA/cuda-kernel-calls.cuf
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ contains
! CHECK: %[[A:.*]]:2 = hlfir.declare %{{.*}} {cuda_attr = #fir.cuda<device>, uniq_name = "_QMtest_callFhostEa"} : (!fir.ref<f32>) -> (!fir.ref<f32>, !fir.ref<f32>)

call dev_kernel0<<<10, 20>>>()
! CHECK: fir.cuda_kernel_launch @_QMtest_callPdev_kernel0<<<%c10{{.*}}, %c1{{.*}}, %c20{{.*}}, %c1{{.*}}, %c1{{.*}}>>>()
! CHECK: fir.cuda_kernel_launch @_QMtest_callPdev_kernel0<<<%c10{{.*}}, %c1{{.*}}, %c1{{.*}}, %c20{{.*}}, %c1{{.*}}, %c1{{.*}}>>>()

call dev_kernel0<<< __builtin_dim3(1,1), __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}>>
Expand All @@ -35,16 +35,16 @@ contains
! CHECK: %[[BLOCK_Y_LOAD:.*]] = fir.load %[[BLOCK_Y]] : !fir.ref<i32>
! CHECK: %[[BLOCK_Z:.*]] = hlfir.designate %[[DIM3_BLOCK]]#1{"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: fir.cuda_kernel_launch @_QMtest_callPdev_kernel0<<<%[[GRID_X_LOAD]], %[[GRID_Y_LOAD]], %[[BLOCK_X_LOAD]], %[[BLOCK_Y_LOAD]], %[[BLOCK_Z_LOAD]]>>>()
! CHECK: fir.cuda_kernel_launch @_QMtest_callPdev_kernel0<<<%[[GRID_X_LOAD]], %[[GRID_Y_LOAD]], %c1{{.*}}, %[[BLOCK_X_LOAD]], %[[BLOCK_Y_LOAD]], %[[BLOCK_Z_LOAD]]>>>()

call dev_kernel0<<<10, 20, 2>>>()
! CHECK: fir.cuda_kernel_launch @_QMtest_callPdev_kernel0<<<%c10{{.*}}, %c1{{.*}}, %c20{{.*}}, %c1{{.*}}, %c1{{.*}}, %c2{{.*}}>>>()
! CHECK: fir.cuda_kernel_launch @_QMtest_callPdev_kernel0<<<%c10{{.*}}, %c1{{.*}}, %c1{{.*}}, %c20{{.*}}, %c1{{.*}}, %c1{{.*}}, %c2{{.*}}>>>()

call dev_kernel0<<<10, 20, 2, 0>>>()
! CHECK: fir.cuda_kernel_launch @_QMtest_callPdev_kernel0<<<%c10{{.*}}, %c1{{.*}}, %c20{{.*}}, %c1{{.*}}, %c1{{.*}}, %c2{{.*}}, %c0{{.*}}>>>()
! CHECK: fir.cuda_kernel_launch @_QMtest_callPdev_kernel0<<<%c10{{.*}}, %c1{{.*}}, %c1{{.*}}, %c20{{.*}}, %c1{{.*}}, %c1{{.*}}, %c2{{.*}}, %c0{{.*}}>>>()

call dev_kernel1<<<1, 32>>>(a)
! CHECK: fir.cuda_kernel_launch @_QMtest_callPdev_kernel1<<<%c1{{.*}}, %c1{{.*}}, %c32{{.*}}, %c1{{.*}}, %c1{{.*}}>>>(%1#1 : !fir.ref<f32>)
! CHECK: fir.cuda_kernel_launch @_QMtest_callPdev_kernel1<<<%c1{{.*}}, %c1{{.*}}, %c1{{.*}}, %c32{{.*}}, %c1{{.*}}, %c1{{.*}}>>>(%1#1 : !fir.ref<f32>)
end

end