Author: abataev
Date: Mon Sep 23 07:06:51 2019
New Revision: 372609
URL: http://llvm.org/viewvc/llvm-project?rev=372609&view=rev
Log:
[OPENMP]Call __kmpc_push_tripcount in task context.
Runtime function __kmpc_push_tripcount better to call inside of the task
context for target regions. Otherwise, the libomptarget is unable to
link the provided tripcount value for nowait target regions and
completely looses this information.
Modified:
cfe/trunk/lib/CodeGen/CGOpenMPRuntime.cpp
cfe/trunk/lib/CodeGen/CGOpenMPRuntime.h
cfe/trunk/lib/CodeGen/CGStmtOpenMP.cpp
cfe/trunk/test/OpenMP/target_teams_distribute_parallel_for_codegen.cpp
cfe/trunk/test/OpenMP/target_teams_distribute_parallel_for_simd_codegen.cpp
cfe/trunk/test/OpenMP/teams_distribute_codegen.cpp
cfe/trunk/test/OpenMP/teams_distribute_parallel_for_codegen.cpp
cfe/trunk/test/OpenMP/teams_distribute_parallel_for_simd_codegen.cpp
cfe/trunk/test/OpenMP/teams_distribute_simd_codegen.cpp
Modified: cfe/trunk/lib/CodeGen/CGOpenMPRuntime.cpp
URL:
http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CGOpenMPRuntime.cpp?rev=372609&r1=372608&r2=372609&view=diff
==============================================================================
--- cfe/trunk/lib/CodeGen/CGOpenMPRuntime.cpp (original)
+++ cfe/trunk/lib/CodeGen/CGOpenMPRuntime.cpp Mon Sep 23 07:06:51 2019
@@ -9176,9 +9176,11 @@ void CGOpenMPRuntime::emitUDMapperArrayI
}
void CGOpenMPRuntime::emitTargetNumIterationsCall(
- CodeGenFunction &CGF, const OMPExecutableDirective &D, const Expr *Device,
- const llvm::function_ref<llvm::Value *(
- CodeGenFunction &CGF, const OMPLoopDirective &D)> &SizeEmitter) {
+ CodeGenFunction &CGF, const OMPExecutableDirective &D,
+ llvm::Value *DeviceID,
+ llvm::function_ref<llvm::Value *(CodeGenFunction &CGF,
+ const OMPLoopDirective &D)>
+ SizeEmitter) {
OpenMPDirectiveKind Kind = D.getDirectiveKind();
const OMPExecutableDirective *TD = &D;
// Get nested teams distribute kind directive, if any.
@@ -9187,30 +9189,24 @@ void CGOpenMPRuntime::emitTargetNumItera
if (!TD)
return;
const auto *LD = cast<OMPLoopDirective>(TD);
- auto &&CodeGen = [LD, &Device, &SizeEmitter, this](CodeGenFunction &CGF,
+ auto &&CodeGen = [LD, DeviceID, SizeEmitter, this](CodeGenFunction &CGF,
PrePostActionTy &) {
- llvm::Value *NumIterations = SizeEmitter(CGF, *LD);
-
- // Emit device ID if any.
- llvm::Value *DeviceID;
- if (Device)
- DeviceID = CGF.Builder.CreateIntCast(CGF.EmitScalarExpr(Device),
- CGF.Int64Ty, /*isSigned=*/true);
- else
- DeviceID = CGF.Builder.getInt64(OMP_DEVICEID_UNDEF);
-
- llvm::Value *Args[] = {DeviceID, NumIterations};
- CGF.EmitRuntimeCall(
- createRuntimeFunction(OMPRTL__kmpc_push_target_tripcount), Args);
+ if (llvm::Value *NumIterations = SizeEmitter(CGF, *LD)) {
+ llvm::Value *Args[] = {DeviceID, NumIterations};
+ CGF.EmitRuntimeCall(
+ createRuntimeFunction(OMPRTL__kmpc_push_target_tripcount), Args);
+ }
};
emitInlinedDirective(CGF, OMPD_unknown, CodeGen);
}
-void CGOpenMPRuntime::emitTargetCall(CodeGenFunction &CGF,
- const OMPExecutableDirective &D,
- llvm::Function *OutlinedFn,
- llvm::Value *OutlinedFnID,
- const Expr *IfCond, const Expr *Device) {
+void CGOpenMPRuntime::emitTargetCall(
+ CodeGenFunction &CGF, const OMPExecutableDirective &D,
+ llvm::Function *OutlinedFn, llvm::Value *OutlinedFnID, const Expr *IfCond,
+ const Expr *Device,
+ llvm::function_ref<llvm::Value *(CodeGenFunction &CGF,
+ const OMPLoopDirective &D)>
+ SizeEmitter) {
if (!CGF.HaveInsertPoint())
return;
@@ -9229,8 +9225,8 @@ void CGOpenMPRuntime::emitTargetCall(Cod
llvm::Value *MapTypesArray = nullptr;
// Fill up the pointer arrays and transfer execution to the device.
auto &&ThenGen = [this, Device, OutlinedFn, OutlinedFnID, &D, &InputInfo,
- &MapTypesArray, &CS, RequiresOuterTask,
- &CapturedVars](CodeGenFunction &CGF, PrePostActionTy &) {
+ &MapTypesArray, &CS, RequiresOuterTask, &CapturedVars,
+ SizeEmitter](CodeGenFunction &CGF, PrePostActionTy &) {
// On top of the arrays that were filled up, the target offloading call
// takes as arguments the device id as well as the host pointer. The host
// pointer is used by the runtime library to identify the current target
@@ -9262,6 +9258,9 @@ void CGOpenMPRuntime::emitTargetCall(Cod
llvm::Value *NumTeams = emitNumTeamsForTargetDirective(CGF, D);
llvm::Value *NumThreads = emitNumThreadsForTargetDirective(CGF, D);
+ // Emit tripcount for the target loop-based directive.
+ emitTargetNumIterationsCall(CGF, D, DeviceID, SizeEmitter);
+
bool HasNowait = D.hasClausesOfKind<OMPNowaitClause>();
// The target region is an outlined function launched by the runtime
// via calls __tgt_target() or __tgt_target_teams().
@@ -11285,12 +11284,13 @@ void CGOpenMPSIMDRuntime::emitTargetOutl
llvm_unreachable("Not supported in SIMD-only mode");
}
-void CGOpenMPSIMDRuntime::emitTargetCall(CodeGenFunction &CGF,
- const OMPExecutableDirective &D,
- llvm::Function *OutlinedFn,
- llvm::Value *OutlinedFnID,
- const Expr *IfCond,
- const Expr *Device) {
+void CGOpenMPSIMDRuntime::emitTargetCall(
+ CodeGenFunction &CGF, const OMPExecutableDirective &D,
+ llvm::Function *OutlinedFn, llvm::Value *OutlinedFnID, const Expr *IfCond,
+ const Expr *Device,
+ llvm::function_ref<llvm::Value *(CodeGenFunction &CGF,
+ const OMPLoopDirective &D)>
+ SizeEmitter) {
llvm_unreachable("Not supported in SIMD-only mode");
}
Modified: cfe/trunk/lib/CodeGen/CGOpenMPRuntime.h
URL:
http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CGOpenMPRuntime.h?rev=372609&r1=372608&r2=372609&view=diff
==============================================================================
--- cfe/trunk/lib/CodeGen/CGOpenMPRuntime.h (original)
+++ cfe/trunk/lib/CodeGen/CGOpenMPRuntime.h Mon Sep 23 07:06:51 2019
@@ -793,6 +793,17 @@ private:
/// default.
virtual unsigned getDefaultFirstprivateAddressSpace() const { return 0; }
+ /// Emit code that pushes the trip count of loops associated with constructs
+ /// 'target teams distribute' and 'teams distribute parallel for'.
+ /// \param SizeEmitter Emits the int64 value for the number of iterations of
+ /// the associated loop.
+ void emitTargetNumIterationsCall(
+ CodeGenFunction &CGF, const OMPExecutableDirective &D,
+ llvm::Value *DeviceID,
+ llvm::function_ref<llvm::Value *(CodeGenFunction &CGF,
+ const OMPLoopDirective &D)>
+ SizeEmitter);
+
public:
explicit CGOpenMPRuntime(CodeGenModule &CGM)
: CGOpenMPRuntime(CGM, ".", ".") {}
@@ -1414,15 +1425,6 @@ public:
bool IsOffloadEntry,
const RegionCodeGenTy &CodeGen);
- /// Emit code that pushes the trip count of loops associated with constructs
- /// 'target teams distribute' and 'teams distribute parallel for'.
- /// \param SizeEmitter Emits the int64 value for the number of iterations of
- /// the associated loop.
- virtual void emitTargetNumIterationsCall(
- CodeGenFunction &CGF, const OMPExecutableDirective &D, const Expr
*Device,
- const llvm::function_ref<llvm::Value *(
- CodeGenFunction &CGF, const OMPLoopDirective &D)> &SizeEmitter);
-
/// Emit the target offloading code associated with \a D. The emitted
/// code attempts offloading the execution to the device, an the event of
/// a failure it executes the host version outlined in \a OutlinedFn.
@@ -1433,11 +1435,15 @@ public:
/// directive, or null if no if clause is used.
/// \param Device Expression evaluated in device clause associated with the
/// target directive, or null if no device clause is used.
- virtual void emitTargetCall(CodeGenFunction &CGF,
- const OMPExecutableDirective &D,
- llvm::Function *OutlinedFn,
- llvm::Value *OutlinedFnID, const Expr *IfCond,
- const Expr *Device);
+ /// \param SizeEmitter Callback to emit number of iterations for loop-based
+ /// directives.
+ virtual void
+ emitTargetCall(CodeGenFunction &CGF, const OMPExecutableDirective &D,
+ llvm::Function *OutlinedFn, llvm::Value *OutlinedFnID,
+ const Expr *IfCond, const Expr *Device,
+ llvm::function_ref<llvm::Value *(CodeGenFunction &CGF,
+ const OMPLoopDirective &D)>
+ SizeEmitter);
/// Emit the target regions enclosed in \a GD function definition or
/// the function itself in case it is a valid device function. Returns true
if
@@ -2117,9 +2123,13 @@ public:
/// directive, or null if no if clause is used.
/// \param Device Expression evaluated in device clause associated with the
/// target directive, or null if no device clause is used.
- void emitTargetCall(CodeGenFunction &CGF, const OMPExecutableDirective &D,
- llvm::Function *OutlinedFn, llvm::Value *OutlinedFnID,
- const Expr *IfCond, const Expr *Device) override;
+ void
+ emitTargetCall(CodeGenFunction &CGF, const OMPExecutableDirective &D,
+ llvm::Function *OutlinedFn, llvm::Value *OutlinedFnID,
+ const Expr *IfCond, const Expr *Device,
+ llvm::function_ref<llvm::Value *(CodeGenFunction &CGF,
+ const OMPLoopDirective &D)>
+ SizeEmitter) override;
/// Emit the target regions enclosed in \a GD function definition or
/// the function itself in case it is a valid device function. Returns true
if
Modified: cfe/trunk/lib/CodeGen/CGStmtOpenMP.cpp
URL:
http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CGStmtOpenMP.cpp?rev=372609&r1=372608&r2=372609&view=diff
==============================================================================
--- cfe/trunk/lib/CodeGen/CGStmtOpenMP.cpp (original)
+++ cfe/trunk/lib/CodeGen/CGStmtOpenMP.cpp Mon Sep 23 07:06:51 2019
@@ -4137,18 +4137,21 @@ static void emitCommonOMPTargetDirective
CGM.getOpenMPRuntime().emitTargetOutlinedFunction(S, ParentName, Fn, FnID,
IsOffloadEntry, CodeGen);
OMPLexicalScope Scope(CGF, S, OMPD_task);
- auto &&SizeEmitter = [](CodeGenFunction &CGF, const OMPLoopDirective &D) {
- OMPLoopScope(CGF, D);
- // Emit calculation of the iterations count.
- llvm::Value *NumIterations = CGF.EmitScalarExpr(D.getNumIterations());
- NumIterations = CGF.Builder.CreateIntCast(NumIterations, CGF.Int64Ty,
- /*isSigned=*/false);
- return NumIterations;
+ auto &&SizeEmitter =
+ [IsOffloadEntry](CodeGenFunction &CGF,
+ const OMPLoopDirective &D) -> llvm::Value * {
+ if (IsOffloadEntry) {
+ OMPLoopScope(CGF, D);
+ // Emit calculation of the iterations count.
+ llvm::Value *NumIterations = CGF.EmitScalarExpr(D.getNumIterations());
+ NumIterations = CGF.Builder.CreateIntCast(NumIterations, CGF.Int64Ty,
+ /*isSigned=*/false);
+ return NumIterations;
+ }
+ return nullptr;
};
- if (IsOffloadEntry)
- CGM.getOpenMPRuntime().emitTargetNumIterationsCall(CGF, S, Device,
- SizeEmitter);
- CGM.getOpenMPRuntime().emitTargetCall(CGF, S, Fn, FnID, IfCond, Device);
+ CGM.getOpenMPRuntime().emitTargetCall(CGF, S, Fn, FnID, IfCond, Device,
+ SizeEmitter);
}
static void emitTargetRegion(CodeGenFunction &CGF, const OMPTargetDirective &S,
Modified: cfe/trunk/test/OpenMP/target_teams_distribute_parallel_for_codegen.cpp
URL:
http://llvm.org/viewvc/llvm-project/cfe/trunk/test/OpenMP/target_teams_distribute_parallel_for_codegen.cpp?rev=372609&r1=372608&r2=372609&view=diff
==============================================================================
--- cfe/trunk/test/OpenMP/target_teams_distribute_parallel_for_codegen.cpp
(original)
+++ cfe/trunk/test/OpenMP/target_teams_distribute_parallel_for_codegen.cpp Mon
Sep 23 07:06:51 2019
@@ -54,17 +54,14 @@ int target_teams_fun(int *g){
// discard capture expressions for te and th
// HCK1: = alloca i32,
// HCK1: = alloca i32,
- // HCK1: = alloca i32,
- // HCK1: = alloca i32,
- // HCK1: = alloca i32,
// HCK1: [[N_CAST:%.+]] = alloca i{{32|64}},
// HCK1: [[TE_CAST:%.+]] = alloca i{{32|64}},
// HCK1: [[TH_CAST:%.+]] = alloca i{{32|64}},
- // HCK1: call void @__kmpc_push_target_tripcount(i64 -1, i64 %{{.+}})
// HCK1: [[N_PAR:%.+]] = load{{.+}}, {{.+}} [[N_CAST]],
// HCK1: [[TE_PAR:%.+]] = load{{.+}}, {{.+}} [[TE_CAST]],
// HCK1: [[TH_PAR:%.+]] = load{{.+}}, {{.+}} [[TH_CAST]],
- // HCK1: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 4, i8**
%{{[^,]+}}, i8** %{{[^,]+}},
+ // HCK1: call void @__kmpc_push_target_tripcount(i64 -1, i64 %{{.+}})
+ // HCK1: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 4, i8**
%{{[^,]+}}, i8** %{{[^,]+}},
// HCK1: call void @[[OFFL1:.+]](i{{32|64}} [[N_PAR]], {{.+}}, i{{32|64}}
[[TE_PAR]], i{{32|64}} [[TH_PAR]])
#pragma omp target teams distribute parallel for num_teams(te),
thread_limit(th)
Modified:
cfe/trunk/test/OpenMP/target_teams_distribute_parallel_for_simd_codegen.cpp
URL:
http://llvm.org/viewvc/llvm-project/cfe/trunk/test/OpenMP/target_teams_distribute_parallel_for_simd_codegen.cpp?rev=372609&r1=372608&r2=372609&view=diff
==============================================================================
--- cfe/trunk/test/OpenMP/target_teams_distribute_parallel_for_simd_codegen.cpp
(original)
+++ cfe/trunk/test/OpenMP/target_teams_distribute_parallel_for_simd_codegen.cpp
Mon Sep 23 07:06:51 2019
@@ -52,18 +52,15 @@ int target_teams_fun(int *g){
// discard capture expressions for te and th
// HCK1: = alloca i32,
// HCK1: = alloca i32,
-// HCK1: = alloca i32,
-// HCK1: = alloca i32,
-// HCK1: = alloca i32,
// HCK1: [[I_CAST:%.+]] = alloca i{{32|64}},
// HCK1: [[N_CAST:%.+]] = alloca i{{32|64}},
// HCK1: [[TE_CAST:%.+]] = alloca i{{32|64}},
// HCK1: [[TH_CAST:%.+]] = alloca i{{32|64}},
-// HCK1: call void @__kmpc_push_target_tripcount(i64 -1, i64 %{{.+}})
// HCK1: [[I_PAR:%.+]] = load{{.+}}, {{.+}} [[I_CAST]],
// HCK1: [[N_PAR:%.+]] = load{{.+}}, {{.+}} [[N_CAST]],
// HCK1: [[TE_PAR:%.+]] = load{{.+}}, {{.+}} [[TE_CAST]],
// HCK1: [[TH_PAR:%.+]] = load{{.+}}, {{.+}} [[TH_CAST]],
+// HCK1: call void @__kmpc_push_target_tripcount(i64 -1, i64 %{{.+}})
// HCK1: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 5, i8**
%{{[^,]+}}, i8** %{{[^,]+}},
// HCK1: call void @[[OFFL1:.+]](i{{32|64}} [[I_PAR]], i{{32|64}} [[N_PAR]],
{{.+}}, i{{32|64}} [[TE_PAR]], i{{32|64}} [[TH_PAR]])
@@ -77,8 +74,7 @@ int target_teams_fun(int *g){
// HCK1: call void @[[OFFL2:.+]](i{{64|32}} %{{.+}})
{{{
#pragma omp target teams distribute parallel for simd is_device_ptr(g)
simdlen(8)
- for(
- int i = 0; i < n; i++) {
+ for(int i = 0; i < n; i++) {
a[i] = g[0];
}
}}}
Modified: cfe/trunk/test/OpenMP/teams_distribute_codegen.cpp
URL:
http://llvm.org/viewvc/llvm-project/cfe/trunk/test/OpenMP/teams_distribute_codegen.cpp?rev=372609&r1=372608&r2=372609&view=diff
==============================================================================
--- cfe/trunk/test/OpenMP/teams_distribute_codegen.cpp (original)
+++ cfe/trunk/test/OpenMP/teams_distribute_codegen.cpp Mon Sep 23 07:06:51 2019
@@ -21,22 +21,19 @@
int a[100];
// CK1: define {{.*}}i32 @{{.+}}teams_argument_globali(
-int teams_argument_global(int n){
+int teams_argument_global(int n) {
int te = n / 128;
int th = 128;
// discard n_addr
// CK1: alloca i32,
// CK1: [[TE:%.+]] = alloca i32,
// CK1: [[TH:%.+]] = alloca i32,
- // CK1: alloca i32,
- // CK1: alloca i32,
- // CK1: alloca i32,
// CK1: [[TE_CAST:%.+]] = alloca i{{32|64}},
// CK1: [[TH_CAST:%.+]] = alloca i{{32|64}},
- // CK1: call void @__kmpc_push_target_tripcount(i64 -1, i64 %{{.+}})
// CK1: [[TE_PAR:%.+]] = load{{.+}}, {{.+}} [[TE_CAST]],
// CK1: [[TH_PAR:%.+]] = load{{.+}}, {{.+}} [[TH_CAST]],
+ // CK1: call void @__kmpc_push_target_tripcount(i64 -1, i64 %{{.+}})
// CK1: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 4, i8**
%{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64*
{{.+}}@{{[^,]+}}, i32 0, i32 0), i32 {{.+}}, i32 {{.+}})
// CK1: call void @[[OFFL1:.+]](i{{32|64}} [[TE_PAR]], i{{32|64}} [[TH_PAR]],
Modified: cfe/trunk/test/OpenMP/teams_distribute_parallel_for_codegen.cpp
URL:
http://llvm.org/viewvc/llvm-project/cfe/trunk/test/OpenMP/teams_distribute_parallel_for_codegen.cpp?rev=372609&r1=372608&r2=372609&view=diff
==============================================================================
--- cfe/trunk/test/OpenMP/teams_distribute_parallel_for_codegen.cpp (original)
+++ cfe/trunk/test/OpenMP/teams_distribute_parallel_for_codegen.cpp Mon Sep 23
07:06:51 2019
@@ -28,14 +28,11 @@ int teams_argument_global(int n){
// CK1: alloca i32,
// CK1: [[TE:%.+]] = alloca i32,
// CK1: [[TH:%.+]] = alloca i32,
- // CK1: alloca i32,
- // CK1: alloca i32,
- // CK1: alloca i32,
// CK1: [[TE_CAST:%.+]] = alloca i{{32|64}},
// CK1: [[TH_CAST:%.+]] = alloca i{{32|64}},
- // CK1: call void @__kmpc_push_target_tripcount(i64 -1, i64 %{{.+}})
// CK1: [[TE_PAR:%.+]] = load{{.+}}, {{.+}} [[TE_CAST]],
// CK1: [[TH_PAR:%.+]] = load{{.+}}, {{.+}} [[TH_CAST]],
+ // CK1: call void @__kmpc_push_target_tripcount(i64 -1, i64 %{{.+}})
// CK1: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 4, i8**
%{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64*
{{.+}}@{{[^,]+}}, i32 0, i32 0), i32 {{.+}}, i32 {{.+}})
// CK1: call void @[[OFFL1:.+]](i{{32|64}} [[TE_PAR]], i{{32|64}} [[TH_PAR]],
Modified: cfe/trunk/test/OpenMP/teams_distribute_parallel_for_simd_codegen.cpp
URL:
http://llvm.org/viewvc/llvm-project/cfe/trunk/test/OpenMP/teams_distribute_parallel_for_simd_codegen.cpp?rev=372609&r1=372608&r2=372609&view=diff
==============================================================================
--- cfe/trunk/test/OpenMP/teams_distribute_parallel_for_simd_codegen.cpp
(original)
+++ cfe/trunk/test/OpenMP/teams_distribute_parallel_for_simd_codegen.cpp Mon
Sep 23 07:06:51 2019
@@ -28,15 +28,12 @@ int teams_argument_global(int n){
// CK1: alloca i32,
// CK1: [[TE:%.+]] = alloca i32,
// CK1: [[TH:%.+]] = alloca i32,
- // CK1: alloca i32,
- // CK1: alloca i32,
- // CK1: alloca i32,
// CK1: [[TE_CAST:%.+]] = alloca i{{32|64}},
// CK1: [[TH_CAST:%.+]] = alloca i{{32|64}},
- // CK1: call void @__kmpc_push_target_tripcount(i64 -1, i64 %{{.+}})
// CK1: [[TE_PAR:%.+]] = load{{.+}}, {{.+}} [[TE_CAST]],
// CK1: [[TH_PAR:%.+]] = load{{.+}}, {{.+}} [[TH_CAST]],
+ // CK1: call void @__kmpc_push_target_tripcount(i64 -1, i64 %{{.+}})
// CK1: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 4, i8**
%{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64*
{{.+}}@{{[^,]+}}, i32 0, i32 0)
// CK1: call void @[[OFFL1:.+]](i{{32|64}} [[TE_PAR]], i{{32|64}} [[TH_PAR]],
Modified: cfe/trunk/test/OpenMP/teams_distribute_simd_codegen.cpp
URL:
http://llvm.org/viewvc/llvm-project/cfe/trunk/test/OpenMP/teams_distribute_simd_codegen.cpp?rev=372609&r1=372608&r2=372609&view=diff
==============================================================================
--- cfe/trunk/test/OpenMP/teams_distribute_simd_codegen.cpp (original)
+++ cfe/trunk/test/OpenMP/teams_distribute_simd_codegen.cpp Mon Sep 23 07:06:51
2019
@@ -30,15 +30,12 @@ int teams_argument_global(int n) {
// CK1: alloca i32,
// CK1: [[TE:%.+]] = alloca i32,
// CK1: [[TH:%.+]] = alloca i32,
- // CK1: alloca i32,
- // CK1: alloca i32,
- // CK1: alloca i32,
// CK1: [[TE_CAST:%.+]] = alloca i{{32|64}},
// CK1: [[TH_CAST:%.+]] = alloca i{{32|64}},
- // CK1: call void @__kmpc_push_target_tripcount(i64 -1, i64 %{{.+}})
// CK1: [[TE_PAR:%.+]] = load{{.+}}, {{.+}} [[TE_CAST]],
// CK1: [[TH_PAR:%.+]] = load{{.+}}, {{.+}} [[TH_CAST]],
+ // CK1: call void @__kmpc_push_target_tripcount(i64 -1, i64 %{{.+}})
// CK1: call i32 @__tgt_target_teams(i64 -1, i8* @{{[^,]+}}, i32 5, i8**
%{{[^,]+}}, i8** %{{[^,]+}}, i{{64|32}}* {{.+}}@{{[^,]+}}, i32 0, i32 0), i64*
{{.+}}@{{[^,]+}}, i32 0, i32 0), i32 {{.+}}, i32 1)
// CK1: call void @[[OFFL1:.+]](i{{32|64}} [[TE_PAR]], i{{32|64}} [[TH_PAR]],
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
