Skip to content

[MLIR] Added check for IsTerminator trait #79317

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 2 commits into from
Jan 26, 2024
Merged

[MLIR] Added check for IsTerminator trait #79317

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
41ff24c
[MLIR] Added check for IsTerminator trait
d-smirnov Jan 24, 2024
bd371f8
Removed obsolete checks
d-smirnov Jan 25, 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
10 changes: 3 additions & 7 deletions mlir/lib/Transforms/RemoveDeadValues.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -594,13 +594,9 @@ void RemoveDeadValues::runOnOperation() {
cleanFuncOp(funcOp, module, la);
} else if (auto regionBranchOp = dyn_cast<RegionBranchOpInterface>(op)) {
cleanRegionBranchOp(regionBranchOp, la);
} else if (op->hasTrait<OpTrait::ReturnLike>()) {
// Nothing to do because this terminator is associated with either a
// function op or a region branch op and gets cleaned when these ops are
// cleaned.
} else if (isa<RegionBranchTerminatorOpInterface>(op)) {
// Nothing to do because this terminator is associated with a region
// branch op and gets cleaned when the latter is cleaned.
} else if (op->hasTrait<::mlir::OpTrait::IsTerminator>()) {
// Nothing to do here because this is a terminator op and it should be
// honored with respect to its parent
Copy link
Collaborator

@joker-eph joker-eph Jan 24, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Seems in line with the cleanSimpleOp comment that states:

/// It is assumed that `op` is simple. Here, a simple op is one which isn't a
/// symbol op, a symbol-user op, a region branch op, a branch op, a region
/// branch terminator op, or return-like.

Now isn't this check making obsolete the two others checks for ReturnLike and RegionBranchTerminatorOpInterface? There two should also be Terminator.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The check seems completely relevant for gpu.termnator op as it declared as:

952 def GPU_TerminatorOp : GPU_Op<"terminator", [HasParent<"LaunchOp">,
953 Pure, Terminator]>,
954 Arguments<(ins)>, Results<(outs)> {

Same for gpu.return op

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Which check are you referring to? The new one?

You may have read it the opposite of what I wrote: I am referring to the existing checks being obsolete after this patch.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sorry, misread you.

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

OK, so: can we remove them?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Removed.

} else if (isa<CallOpInterface>(op)) {
// Nothing to do because this op is associated with a function op and gets
// cleaned when the latter is cleaned.
Expand Down
22 changes: 22 additions & 0 deletions mlir/test/Transforms/remove-dead-values.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -335,3 +335,25 @@ func.func @main(%arg3 : i32, %arg4 : i1) {
%non_live_0 = func.call @clean_region_branch_op_erase_it(%arg3, %arg4) : (i32, i1) -> (i32)
return
}

// -----

#map = affine_map<(d0)[s0, s1] -> (d0 * s0 + s1)>
func.func @kernel(%arg0: memref<18xf32>) {
%c1 = arith.constant 1 : index
%c18 = arith.constant 18 : index
gpu.launch blocks(%arg3, %arg4, %arg5) in (%arg9 = %c18, %arg10 = %c18, %arg11 = %c18) threads(%arg6, %arg7, %arg8) in (%arg12 = %c1, %arg13 = %c1, %arg14 = %c1) {
%c1_0 = arith.constant 1 : index
%c0_1 = arith.constant 0 : index
%cst_2 = arith.constant 25.4669495 : f32
%6 = affine.apply #map(%arg3)[%c1_0, %c0_1]
memref.store %cst_2, %arg0[%6] : memref<18xf32>
gpu.terminator
} {SCFToGPU_visited}
return
}

// CHECK-LABEL: func.func @kernel(%arg0: memref<18xf32>) {
// CHECK: gpu.launch blocks
// CHECK: memref.store
// CHECK-NEXT: gpu.terminator