Skip to content

[OpenMP][mlir] Added num_teams, thread_limit translation to LLVM IR #68821

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
Oct 15, 2023
Merged

[OpenMP][mlir] Added num_teams, thread_limit translation to LLVM IR #68821

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
fff04e5
[OpenMP][mlir] Added `num_teams`, `thread_limit` translation to LLVM IR
shraiysh Oct 11, 2023
16a7189
Merge branch 'llvm:main' into teams_clauses1
shraiysh Oct 15, 2023
6811ac6
Address comments
shraiysh Oct 15, 2023
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
22 changes: 16 additions & 6 deletions mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -666,11 +666,9 @@ convertOmpTeams(omp::TeamsOp op, llvm::IRBuilderBase &builder,
LLVM::ModuleTranslation &moduleTranslation) {
using InsertPointTy = llvm::OpenMPIRBuilder::InsertPointTy;
LogicalResult bodyGenStatus = success();
if (op.getNumTeamsLower() || op.getNumTeamsUpper() || op.getIfExpr() ||
op.getThreadLimit() || !op.getAllocatorsVars().empty() ||
op.getReductions()) {
if (op.getIfExpr() || !op.getAllocatorsVars().empty() || op.getReductions())
return op.emitError("unhandled clauses for translation to LLVM IR");
}

auto bodyCB = [&](InsertPointTy allocaIP, InsertPointTy codegenIP) {
LLVM::ModuleTranslation::SaveStack<OpenMPAllocaStackFrame> frame(
moduleTranslation, allocaIP);
Expand All @@ -679,9 +677,21 @@ convertOmpTeams(omp::TeamsOp op, llvm::IRBuilderBase &builder,
moduleTranslation, bodyGenStatus);
};

llvm::Value *numTeamsLower = nullptr;
if (Value numTeamsLowerVar = op.getNumTeamsLower())
numTeamsLower = moduleTranslation.lookupValue(numTeamsLowerVar);

llvm::Value *numTeamsUpper = nullptr;
if (Value numTeamsUpperVar = op.getNumTeamsUpper())
numTeamsUpper = moduleTranslation.lookupValue(numTeamsUpperVar);

llvm::Value *threadLimit = nullptr;
if (Value threadLimitVar = op.getThreadLimit())
threadLimit = moduleTranslation.lookupValue(threadLimitVar);

llvm::OpenMPIRBuilder::LocationDescription ompLoc(builder);
builder.restoreIP(
moduleTranslation.getOpenMPBuilder()->createTeams(ompLoc, bodyCB));
builder.restoreIP(moduleTranslation.getOpenMPBuilder()->createTeams(
ompLoc, bodyCB, numTeamsLower, numTeamsUpper, threadLimit));
return bodyGenStatus;
}

Expand Down
111 changes: 111 additions & 0 deletions mlir/test/Target/LLVMIR/openmp-teams.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -124,3 +124,114 @@ llvm.func @omp_teams_branching_shared(%condition: i1, %arg0: i32, %arg1: f32, %a
// CHECK-NEXT: br label
// CHECK: ret void

// -----

llvm.func @beforeTeams()
llvm.func @duringTeams()
llvm.func @afterTeams()

// CHECK-LABEL: @omp_teams_thread_limit
// CHECK-SAME: (i32 [[THREAD_LIMIT:.+]])
llvm.func @omp_teams_thread_limit(%threadLimit: i32) {
// CHECK-NEXT: call void @beforeTeams()
llvm.call @beforeTeams() : () -> ()
// CHECK: [[THREAD_NUM:%.+]] = call i32 @__kmpc_global_thread_num
// CHECK-NEXT: call void @__kmpc_push_num_teams_51({{.+}}, i32 [[THREAD_NUM]], i32 0, i32 0, i32 [[THREAD_LIMIT]])
// CHECK: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @1, i32 0, ptr [[OUTLINED_FN:.+]])
omp.teams thread_limit(%threadLimit : i32) {
llvm.call @duringTeams() : () -> ()
omp.terminator
}
// CHECK: call void @afterTeams
llvm.call @afterTeams() : () -> ()
// CHECK: ret void
llvm.return
}

// CHECK: define internal void [[OUTLINED_FN]](ptr {{.+}}, ptr {{.+}})
// CHECK: call void @duringTeams()
// CHECK: ret void

// -----

llvm.func @beforeTeams()
llvm.func @duringTeams()
llvm.func @afterTeams()

// CHECK-LABEL: @omp_teams_num_teams_upper
// CHECK-SAME: (i32 [[NUM_TEAMS_UPPER:.+]])
llvm.func @omp_teams_num_teams_upper(%numTeamsUpper: i32) {
// CHECK-NEXT: call void @beforeTeams()
llvm.call @beforeTeams() : () -> ()
// CHECK: [[THREAD_NUM:%.+]] = call i32 @__kmpc_global_thread_num
// CHECK-NEXT: call void @__kmpc_push_num_teams_51({{.+}}, i32 [[THREAD_NUM]], i32 [[NUM_TEAMS_UPPER]], i32 [[NUM_TEAMS_UPPER]], i32 0)
// CHECK: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @1, i32 0, ptr [[OUTLINED_FN:.+]])
omp.teams num_teams(to %numTeamsUpper : i32) {
llvm.call @duringTeams() : () -> ()
omp.terminator
}
// CHECK: call void @afterTeams
llvm.call @afterTeams() : () -> ()
// CHECK: ret void
llvm.return
}

// CHECK: define internal void [[OUTLINED_FN]](ptr {{.+}}, ptr {{.+}})
// CHECK: call void @duringTeams()
// CHECK: ret void

// -----

llvm.func @beforeTeams()
llvm.func @duringTeams()
llvm.func @afterTeams()

// CHECK-LABEL: @omp_teams_num_teams_lower_and_upper
// CHECK-SAME: (i32 [[NUM_TEAMS_LOWER:.+]], i32 [[NUM_TEAMS_UPPER:.+]])
llvm.func @omp_teams_num_teams_lower_and_upper(%numTeamsLower: i32, %numTeamsUpper: i32) {
// CHECK-NEXT: call void @beforeTeams()
llvm.call @beforeTeams() : () -> ()
// CHECK: [[THREAD_NUM:%.+]] = call i32 @__kmpc_global_thread_num
// CHECK-NEXT: call void @__kmpc_push_num_teams_51({{.+}}, i32 [[THREAD_NUM]], i32 [[NUM_TEAMS_LOWER]], i32 [[NUM_TEAMS_UPPER]], i32 0)
// CHECK: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @1, i32 0, ptr [[OUTLINED_FN:.+]])
omp.teams num_teams(%numTeamsLower : i32 to %numTeamsUpper: i32) {
llvm.call @duringTeams() : () -> ()
omp.terminator
}
// CHECK: call void @afterTeams
llvm.call @afterTeams() : () -> ()
// CHECK: ret void
llvm.return
}

// CHECK: define internal void [[OUTLINED_FN]](ptr {{.+}}, ptr {{.+}})
// CHECK: call void @duringTeams()
// CHECK: ret void

// -----

llvm.func @beforeTeams()
llvm.func @duringTeams()
llvm.func @afterTeams()

// CHECK-LABEL: @omp_teams_num_teams_and_thread_limit
// CHECK-SAME: (i32 [[NUM_TEAMS_LOWER:.+]], i32 [[NUM_TEAMS_UPPER:.+]], i32 [[THREAD_LIMIT:.+]])
llvm.func @omp_teams_num_teams_and_thread_limit(%numTeamsLower: i32, %numTeamsUpper: i32, %threadLimit: i32) {
// CHECK-NEXT: call void @beforeTeams()
llvm.call @beforeTeams() : () -> ()
// CHECK: [[THREAD_NUM:%.+]] = call i32 @__kmpc_global_thread_num
// CHECK-NEXT: call void @__kmpc_push_num_teams_51({{.+}}, i32 [[THREAD_NUM]], i32 [[NUM_TEAMS_LOWER]], i32 [[NUM_TEAMS_UPPER]], i32 [[THREAD_LIMIT]])
// CHECK: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @1, i32 0, ptr [[OUTLINED_FN:.+]])
omp.teams num_teams(%numTeamsLower : i32 to %numTeamsUpper: i32) thread_limit(%threadLimit: i32) {
llvm.call @duringTeams() : () -> ()
omp.terminator
}
// CHECK: call void @afterTeams
llvm.call @afterTeams() : () -> ()
// CHECK: ret void
llvm.return
}

// CHECK: define internal void [[OUTLINED_FN]](ptr {{.+}}, ptr {{.+}})
// CHECK: call void @duringTeams()
// CHECK: ret void