[AbstractAttributor] Fold function calls to `__kmpc_is_spmd_exec_mode` if possible

[tianshilei1992]

Summary

In the device runtime there are many function calls to __kmpc_is_spmd_exec_mode
to query the execution mode of current kernels. In many cases, user programs
only contain target region executing in one mode. As a consequence, those runtime
function calls will only return one value. If we can get rid of these function
calls during compliation, it can potentially improve performance.

In this patch, we use AAKernelInfo to analyze kernel execution. Basically, for
each kernel (device) function F, we collect all kernel entries K that can
reach F. In each iteration, we go through all reaching kernel entries and check
their execution mode. If F can only be reached by kernel entries with same mode,
we update a map from CallBase * to Constant * to corresponding value. In
manifest stage, if any entry of the map is not nullptr, which means the function
call can be folded to the Constant *, we replace all uses of the function call
and remove it.

Later we will also add more foldable functions, such as isMainThread.

[jdoerfert]

Test missing. Other than that mostly nits.

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

460

Unsure, I think this way is more natural.

464

2738

use the changeValueAfterManifest interface.

3007–3008

Remove

3025

3105

Move this before the kernel handling, then you can exit if it is not a kernel. Alternatively, put the kernel handling in a helper.

3152–3153

3202

Maybe put this into a helper as well. And the "reaching kernels" update. Will make it easier to read in the future.

3209

Add a comment why we would want this (here).

[tianshilei1992]

fix coments

[JonChesterfield]

Interesting, thanks. For the functions that could be called by generic or spmd kernels, we could clone them (one for each) as (I think) we know at the entry point which it will be.

We probably want this after any transforms that try to convert generic kernels to spmd. Not sure how to order that.

[jdoerfert]

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

2735–2738

[tianshilei1992]

• fixed comments
• added a wrapper for kernel entry set up

[tianshilei1992]

rebase

[tianshilei1992]

fixed rebase issue

[tianshilei1992]

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

3316

This condition should include SPMDization, but what should be used?

[tianshilei1992]

rebase

[tianshilei1992]

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

458

We probably don't need this variable at all.

[tianshilei1992]

rebase

[jdoerfert]

Test missing.

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

492–493

?

2730–2745

here we need to return changed or we need to set changed to CHANGED if changeToSPMDMode returned true.

3141

Why !assumed() ? Wouldn't that mean we only update if we have unknown reaching kernels?

3266

Early exit please.

3275

I'm not so sure about this assertion. Maybe handle the case conservatively instead.

3276–3277

[tianshilei1992]

fixed comments

[tianshilei1992]

revmoed unrelated change

[jdoerfert]

Test still missing ;)

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

3141

This is better but I'm still not sure if it's correct. When we go from valid to invalid we stop updating the map, which is bad because the content is still in there.

[JonChesterfield]

I wonder if this is separable.

Could we have an attribute, 'openmp_called_in_spmd_mode' or similar, which we apply to functions when that is always true, similar to the above logic.

We could then have a pass which, for functions with that attribute set, folds call instructions within that function appropriately.

edit: Partly thinking of splitting it to simplify testing - we can write various call trees and check the right ones pick up the attribute, and separate tests to check we fold the calls based on that attribute.

[tianshilei1992]

added tests

[tianshilei1992]

llvm/test/Transforms/OpenMP/is_spmd_exec_mode_unfoldable.ll

343 ↗

(On Diff #357330)

This function will not be called if we remove the way to tell if it is in SPMD mode from ident.

[tianshilei1992]

update foldable test

[tianshilei1992]

update the test case with update_test_checks

[tianshilei1992]

update test

[JonChesterfield]

Thought about this a little more. I think there's a dead-simple 90% solution available.

-For each internal global

• If it is undef initialized
• If it is only used by loads and stores (thus no aliasing)
• If every store writes the same constant
• Replace every load with that constant, then delete the stores, then delete the global

That's a very narrow load/store propagation that only applies to globals that are initially undef, which are unusual in most programs, but common where __shared__ or pteam memory is involved.

If I'm reading some disassembled IR right, it would work for the variables used to report is_spmd and is_uninitialized. For modules that only contain one type of kernel, the overall effect would be the same as the current patch, at much reduced complexity.

Should I type that up? I think we're only looking to reduce the cost of the current/legacy devicertl with this patch. When more serious aliasing analysis comes online it would obsolete the trivial IR transform above, at which point we can delete said transform and keep the verification tests. I particularly like the path to cleanly deleting the pass when it becomes obsolete.

edit: Tried to write a test case for this, but everything I've tried so far gets stripped out by the optimiser already. i.e. the is_spmd_exec_mode calls are already removed. Can you share the test case/application where this is not the case?

[jdoerfert]

In D105246#2866459, @JonChesterfield wrote:

Thought about this a little more. I think there's a dead-simple 90% solution available.

-For each internal global

• If it is undef initialized
• If it is only used by loads and stores (thus no aliasing)
• If every store writes the same constant
• Replace every load with that constant, then delete the stores, then delete the global

And if you make it "field-sensitive" now you are almost at https://reviews.llvm.org/D104432 :)
Not sure I'd call that dead-simple but I'm with you.

That's a very narrow load/store propagation that only applies to globals that are initially undef, which are unusual in most programs, but common where __shared__ or pteam memory is involved.

If I'm reading some disassembled IR right, it would work for the variables used to report is_spmd and is_uninitialized. For modules that only contain one type of kernel, the overall effect would be the same as the current patch, at much reduced complexity.

Should I type that up? I think we're only looking to reduce the cost of the current/legacy devicertl with this patch. When more serious aliasing analysis comes online it would obsolete the trivial IR transform above, at which point we can delete said transform and keep the verification tests. I particularly like the path to cleanly deleting the pass when it becomes obsolete.

edit: Tried to write a test case for this, but everything I've tried so far gets stripped out by the optimiser already. i.e. the is_spmd_exec_mode calls are already removed. Can you share the test case/application where this is not the case?

[tianshilei1992]

Will move the change out of AAKernelInfo.

[tianshilei1992]

Move to D105787

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

struct KernelInfoState : AbstractState {
   /// The __kmpc_target_init call in this kernel, if any. If we find more than
   /// one we abort as the kernel is malformed.
   CallBase *KernelInitCB = nullptr;
 
   /// The __kmpc_target_deinit call in this kernel, if any. If we find more than
   /// one we abort as the kernel is malformed.
   CallBase *KernelDeinitCB = nullptr;
 
+  /// A map from a function to its constant return value. If the value is
+  /// nullptr, the function cannot be folded.
+  SmallDenseMap<CallBase *, Constant *> FoldableFunctions;
+
   /// Flag to indicate that we may reach a parallel region that is not tracked
   /// in the ParallelRegions set above.
   SmallPtrSet<CallBase *, 4> UnknownParallelRegions;
 
+  /// Flag to indicate that we may reach a kernel entry that is not tracked.
+  bool MayBeReachedByUnknownKernel = false;
+
   /// Flag to indicate if the associated function can be executed in SPMD mode.
+  /// Note that it could be set to false if the function is already in SPMD
+  /// mode.
   bool IsSPMDCompatible = true;
 
+  /// Flag to indicate if the associated function is in SPMD mode. This is
+  /// different from \p IsSPMDCompatible. This variable is only set to true if
+  /// it is indeed in a SPMD mode, not from being spmdized.
+  bool IsSPMDMode = false;
+
+  /// Flag to indicate if the associated function is a kernel entry.
+  bool IsKernelEntry = false;
+
+  /// Kernels that can reach the associated function.
+  SmallPtrSet<Kernel, 4> ReachingKernels;
+
   /// Abstract State interface
   ///{
 
   KernelInfoState() {}
   KernelInfoState(bool BestState) {
     if (!BestState)
       indicatePessimisticFixpoint();
   }
 
   /// See AbstractState::isValidState(...)
   bool isValidState() const override { return true; }
 
   /// See AbstractState::isAtFixpoint(...)
   bool isAtFixpoint() const override { return IsAtFixpoint; }
 
   /// See AbstractState::indicatePessimisticFixpoint(...)
   ChangeStatus indicatePessimisticFixpoint() override {
     IsAtFixpoint = true;
     IsSPMDCompatible = false;
     UnknownParallelRegions.insert(nullptr);
+    MayBeReachedByUnknownKernel = true;
     return ChangeStatus::CHANGED;
   }
 
   /// See AbstractState::indicateOptimisticFixpoint(...)
   ChangeStatus indicateOptimisticFixpoint() override {
     IsAtFixpoint = true;
     return ChangeStatus::UNCHANGED;
   }
 
   /// Return the assumed state
   KernelInfoState &getAssumed() { return *this; }
   const KernelInfoState &getAssumed() const { return *this; }
 
   bool operator==(const KernelInfoState &RHS) const {
     if ((UnknownParallelRegions != RHS.UnknownParallelRegions) ||
         (IsSPMDCompatible != RHS.IsSPMDCompatible))
       return false;
-    return ParallelRegions == RHS.ParallelRegions;
+
+    if (ParallelRegions != RHS.ParallelRegions)
+      return false;
+
+    if (ReachingKernels != RHS.ReachingKernels)
+      return false;
+
+    if (FoldableFunctions != RHS.FoldableFunctions)
+      return false;
+
+    return true;
   }
 
   /// Return empty set as the best state of potential values.
   static KernelInfoState getBestState() { return KernelInfoState(true); }
 
   static KernelInfoState getBestState(KernelInfoState &KIS) {
     return getBestState();
   }
KernelInfoState operator^=(const KernelInfoState &KIS) {
     }
     if (KIS.KernelDeinitCB) {
       if (KernelDeinitCB && KernelDeinitCB != KIS.KernelDeinitCB)
         indicatePessimisticFixpoint();
       KernelDeinitCB = KIS.KernelDeinitCB;
     }
     UnknownParallelRegions.insert(KIS.UnknownParallelRegions.begin(),
                                   KIS.UnknownParallelRegions.end());
+    MayBeReachedByUnknownKernel |= KIS.MayBeReachedByUnknownKernel;
     IsSPMDCompatible &= KIS.IsSPMDCompatible;
     ParallelRegions.insert(KIS.ParallelRegions.begin(),
                            KIS.ParallelRegions.end());
     return *this;
   }
 
   KernelInfoState operator&=(const KernelInfoState &KIS) {
     return (*this ^= KIS);
InitRFI.foreachUse(
           return false;
         },
         getAnchorScope());
   }
 
   /// Modify the IR based on the KernelInfoState as the fixpoint iteration is
   /// finished now.
   ChangeStatus manifest(Attributor &A) override {
+    ChangeStatus Change = ChangeStatus::UNCHANGED;
+
     // This is a somewhat unrelated modification. We basically flatten the
     // function that was reached from a kernel completely by asking the inliner
     // to inline everything it can. This should live in a separate AA though as
     // it should also run on parallel regions and other GPU functions.
     auto CheckCallInst = [&](Instruction &I) {
       auto &CB = cast<CallBase>(I);
       CB.addAttribute(AttributeList::FunctionIndex, Attribute::AlwaysInline);
       return true;
     };
     A.checkForAllCallLikeInstructions(CheckCallInst, *this);
 
-    return buildCustomStateMachine(A);
+    // Fold all valid foldable functions
+    for (std::pair<CallBase *, Constant *> &P : FoldableFunctions) {
+      if (P.second == nullptr)
+        continue;
+
+      for (Use &U : P.first->uses()) {
+        A.changeUseAfterManifest(U, *P.second);
+        A.deleteAfterManifest(*P.first);
+      }
+
+      Change = ChangeStatus::CHANGED;
+    }
+
+    Change = Change | buildCustomStateMachine(A);
+
+    return Change;
   }
 
   ChangeStatus buildCustomStateMachine(Attributor &A) {
     // If we are not looking at a kernel with __kmpc_target_init and
     // __kmpc_target_deinit call we cannot actually manifest the information.
     if (!KernelInitCB || !KernelDeinitCB)
       return ChangeStatus::UNCHANGED;
 
ChangeStatus buildCustomStateMachine(Attributor &A) {
     BranchInst::Create(StateMachineDoneBarrierBB, StateMachineEndParallelBB);
 
     CallInst::Create(BarrierFn, {Ident, GTid}, "", StateMachineDoneBarrierBB);
     BranchInst::Create(StateMachineBeginBB, StateMachineDoneBarrierBB);
 
     return ChangeStatus::CHANGED;
   }
 
   /// Returns true if value is assumed to be tracked.
   bool canBeExecutedInSPMDMode() const { return IsSPMDCompatible; }
 
+  /// TODO: This function might be redundant if \p IsSPMDCompatible is NOT set
+  /// to false if the function is already in SPMD mode.
+  /// Returns true if the associated function will be executed in SPMD mode, no
+  /// matter whether it is initially in SPMD mode or spmdized.
+  bool willBeExecutedInSPMDMode() const {
+    return IsSPMDMode || IsSPMDCompatible;
+  }
+
   /// Statistics are tracked as part of manifest for now.
   void trackStatistics() const override {}
 
   /// See AbstractAttribute::getAsStr()
   const std::string getAsStr() const override {
     if (!isValidState())
       return "<invalid>";
-    return std::string(canBeExecutedInSPMDMode() ? "SPMD" : "generic") + std::string("#PRs: ") + std::to_string(ParallelRegions.size()) +
+    return std::string(canBeExecutedInSPMDMode() ? "SPMD" : "generic") +
+           std::string("#PRs: ") + std::to_string(ParallelRegions.size()) +
            ", #Unknown PRs: " + std::to_string(!UnknownParallelRegions.size());
-||||||| parent of 3665f4d7e56f (Manually rebase D102307, no tests)
   }
 
   /// Create an abstract attribute biew for the position \p IRP.
   static AAKernelInfo &createForPosition(const IRPosition &IRP, Attributor &A);
 
   /// See AbstractAttribute::getName()
   const std::string getName() const override { return "AAKernelInfo"; }
 

 };
 
 /// The function kernel info abstract attribute, basically, what can we say
 /// about a function with regards to the KernelInfoState.
 struct AAKernelInfoFunction : AAKernelInfo {
   AAKernelInfoFunction(const IRPosition &IRP, Attributor &A)
       : AAKernelInfo(IRP, A) {}
 
+  void initialize(Attributor &A) override {
+    AAKernelInfo::initialize(A);
+
+    Function *F = getAssociatedFunction();
+
+    // If the attached function is a kernel, we simply add itself to the set of
+    // reached kernels.
+    auto &OMPInfoCache = static_cast<OMPInformationCache &>(A.getInfoCache());
+    if (OMPInfoCache.Kernels.count(F)) {
+      ReachingKernels.insert(F);
+      IsKernelEntry = true;
+      CallBase *CB = nullptr;
+
+      OMPInformationCache::RuntimeFunctionInfo &TargetInitRFI =
+          OMPInfoCache.RFIs[OMPRTL___kmpc_target_init];
+      TargetInitRFI.foreachUse(
+          [&](Use &U, Function &Caller) {
+            assert(CB == nullptr && "__kmpc_target_init is called more than "
+                                    "once in one kernel entry");
+
+            assert(dyn_cast<CallBase>(U.getUser()) &&
+                   "Use of __kmpc_target_init is not a CallBase");
+
+            CB = cast<CallBase>(U.getUser());
+
+            return true;
+          },
+          F);
+
+      assert(CB && "Call to __kmpc_target_init is missing");
+
+      constexpr const int InitIsSPMDArgNo = 1;
+
+      ConstantInt *IsSPMDArg =
+          dyn_cast<ConstantInt>(CB->getArgOperand(InitIsSPMDArgNo));
+
+      if (IsSPMDArg && !IsSPMDArg->isZero())
+        IsSPMDMode = true;
+    }
+
+    // For each use of __kmpc_is_spmd_exec_mode, we set up the FoldableFunctions
+    // based on its current SPMD mode.
+    OMPInformationCache::RuntimeFunctionInfo &IsSPMDExecModeRFI =
+        OMPInfoCache.RFIs[OMPRTL___kmpc_is_spmd_exec_mode];
+    IsSPMDExecModeRFI.foreachUse(
+        [&](Use &U, Function &Caller) {
+          CallBase *CB = dyn_cast<CallBase>(U.getUser());
+          assert(CB && "Use of __kmpc_is_spmd_exec_mode is not a CallBase");
+          FoldableFunctions[CB] =
+              ConstantInt::get(Type::getInt8Ty(Caller.getContext()), 1);
+          return true;
+        },
+        F);
+  }
+
   /// Fixpoint iteration update function. Will be called every time a dependence
   /// changed its state (and in the beginning).
   ChangeStatus updateImpl(Attributor &A) override {
     KernelInfoState StateBefore = getState();
 
     // Callback to check a read/write instruction.
     auto CheckRWInst = [&](Instruction &I) {
       // We handle calls later.
auto CheckRWInst = [&](Instruction &I) {
       // For now we give up on everything but stores.
       IsSPMDCompatible = false;
       return true;
     };
     if (IsSPMDCompatible &&
         !A.checkForAllReadWriteInstructions(CheckRWInst, *this))
       IsSPMDCompatible = false;
 
+
+
+    if (!IsKernelEntry) {
+      auto PredCallSite = [&](AbstractCallSite ACS) {
+        Function *Caller = ACS.getInstruction()->getFunction();
+
+        assert(Caller && "Caller is nullptr");
+
+        auto &CAA =
+            A.getOrCreateAAFor<AAKernelInfo>(IRPosition::function(*Caller));
+        if (CAA.isValidState()) {
+          ReachingKernels.insert(CAA.ReachingKernels.begin(),
+                                 CAA.ReachingKernels.end());
+          return true;
+        }
+
+        // We might encounter an unknown kernel. We cannot fold the associated
+        // function here.
+        MayBeReachedByUnknownKernel = true;
+
+        return true;
+      };
+
+      bool AllCallSitesKnown;
+      if (!A.checkForAllCallSites(PredCallSite, *this,
+                                  true /* RequireAllCallSites */,
+                                  AllCallSitesKnown))
+        MayBeReachedByUnknownKernel = true;
+    }
+
+    if (!MayBeReachedByUnknownKernel) {
+      unsigned Count = 0;
+
+      for (Kernel K : ReachingKernels) {
+        auto &AA = A.getAAFor<AAKernelInfo>(*this, IRPosition::function(*K),
+                                            DepClassTy::REQUIRED);
+        assert(AA.isValidState() && "AA should be valid here");
+        if (AA.willBeExecutedInSPMDMode())
+          ++Count;
+      }
+
+      // Assume reaching kernels are in a mixture of SPMD and non-SPMD mode.
+      // Update all function calls to __kmpc_is_spmd_exec_mode to nullptr.
+      Constant *C = nullptr;
+
+      auto &Ctx = getAnchorValue().getContext();
+
+      if (Count == 0) {
+        // All reaching kernels are in non-SPMD mode. Update all function
+        // calls to __kmpc_is_spmd_exec_mode to 0.
+        C = ConstantInt::get(Type::getInt8Ty(Ctx), 0);
+      } else if (Count == ReachingKernels.size()) {
+        // All reaching kernels are in SPMD mode. Update all function calls to
+        // __kmpc_is_spmd_exec_mode to 1.
+        C = ConstantInt::get(Type::getInt8Ty(Ctx), 1);
+      }
+
+      auto &OMPInfoCache = static_cast<OMPInformationCache &>(A.getInfoCache());
+      OMPInformationCache::RuntimeFunctionInfo &IsSPMDExecModeRFI =
+          OMPInfoCache.RFIs[OMPRTL___kmpc_is_spmd_exec_mode];
+
+      for (std::pair<CallBase *, Constant *> &P : FoldableFunctions) {
+        CallBase *CB = P.first;
+        if (CB->getCalledFunction() == IsSPMDExecModeRFI.Declaration)
+          P.second = C;
+      }
+    }
+
     // Callback to check a call instruction.
     auto CheckCallInst = [&](Instruction &I) {
       auto &CB = cast<CallBase>(I);
       Function *Callee = CB.getCalledFunction();
       if (Callee) {
+        A.getOrCreateAAFor<AAKernelInfo>(IRPosition::function(*Callee));
+
         auto &CBAA = A.getAAFor<AAKernelInfo>(
             *this, IRPosition::callsite_function(CB), DepClassTy::REQUIRED);
         if (CBAA.getState().isValidState()) {
           getState() ^= CBAA.getState();
           return true;
         }
       }
 
if (!CB.mayWriteToMemory() || isa<IntrinsicInst>(CB) ||
       indicateOptimisticFixpoint();
       return;
     }
 
     // Next we check if we know the callee. If it is a known OpenMP function
     // we will handle them explicitly in the switch below. If it is not, we
     // will use an AAKernelInfo object on the callee to gather information and
     // merge that into the current state. The latter happens in the updateImpl.
     auto &OMPInfoCache = static_cast<OMPInformationCache &>(A.getInfoCache());
     const auto &It = OMPInfoCache.RuntimeFunctionIDMap.find(Callee);
     if (It == OMPInfoCache.RuntimeFunctionIDMap.end()) {
       // Unknown caller or declarations are not analyzable, we give up.
       if (!Callee || !A.isFunctionIPOAmendable(*Callee))
         indicatePessimisticFixpoint();
       return;
     }
 
     const unsigned int WrapperFunctionArgNo = 6;
     RuntimeFunction RF = It->getSecond();
     switch (RF) {
     // All the functions we know are compatible with SPMD mode.
     case OMPRTL___kmpc_for_static_fini:
     case OMPRTL___kmpc_global_thread_num:
     case OMPRTL___kmpc_single:
     case OMPRTL___kmpc_end_single:
     case OMPRTL___kmpc_master:
     case OMPRTL___kmpc_end_master:
     case OMPRTL___kmpc_barrier:
void initialize(Attributor &A) override {
     case OMPRTL___kmpc_target_init:
       KernelInitCB = &CB;
       break;
     case OMPRTL___kmpc_target_deinit:
       KernelDeinitCB = &CB;
       break;
     case OMPRTL___kmpc_parallel_51:
       if (auto *ParallelRegion = dyn_cast<Function>(
               CB.getArgOperand(WrapperFunctionArgNo)->stripPointerCasts())) {
         ParallelRegions.insert(ParallelRegion);
         break;
       }
       // The condition above should usually get the parallel region function
       // pointer and record it. In the off chance it doesn't we assume the
       // worst.
       UnknownParallelRegions.insert(&CB);
       break;