[PATCH 15/38] [noalias] D9382: llvm.noalias - don't prevent loop vectorization

[jeroen.dobbelaere]

Summary

This is part of the series started by D68484.

Rebase D9382

[lebedev.ri]

llvm/test/Transforms/LoopVectorize/noalias.ll

6

s/lifetime/aliasing/ ?

8–10

It might be good to check a bit more context - *how* did it get vectorized?

[a.elovikov]

llvm/test/Transforms/LoopVectorize/noalias.ll

8–10

And, in addition to this, it might be good to see how it's vectorized for the interleaved accesses (especially how the intrinsics/shuffles "interaction" looks like), e.g (similar to interleaved-accesses-1.ll).

struct S {
  int * restrict i;
  float * restrict f;
};

S *p1, *p2;
for (int j = 0; j < N; ++j) {
  S t = p1[j];
  p2[j] = t;
}

[jeroen.dobbelaere]

Rebased

[jeroen.dobbelaere]

For review convenience, all changes related to the original local restrict patches have now been incorporated in the full restrict patch series.

llvm/include/llvm/Analysis/TargetTransformInfoImpl.h

unsigned getIntrinsicCost(Intrinsic::ID IID, Type *RetTy,
     case Intrinsic::dbg_label:
     case Intrinsic::invariant_start:
     case Intrinsic::invariant_end:
     case Intrinsic::launder_invariant_group:
     case Intrinsic::strip_invariant_group:
     case Intrinsic::is_constant:
     case Intrinsic::lifetime_start:
     case Intrinsic::lifetime_end:
+    case Intrinsic::noalias:
     case Intrinsic::objectsize:
     case Intrinsic::ptr_annotation:
     case Intrinsic::var_annotation:
     case Intrinsic::experimental_gc_result:
     case Intrinsic::experimental_gc_relocate:
     case Intrinsic::coro_alloc:
     case Intrinsic::coro_begin:
     case Intrinsic::coro_free:

llvm/lib/Analysis/VectorUtils.cpp

 Intrinsic::ID llvm::getVectorIntrinsicIDForCall(const CallInst *CI,
                                                 const TargetLibraryInfo *TLI) {
   Intrinsic::ID ID = getIntrinsicForCallSite(CI, TLI);
   if (ID == Intrinsic::not_intrinsic)
     return Intrinsic::not_intrinsic;
 
   if (isTriviallyVectorizable(ID) || ID == Intrinsic::lifetime_start ||
       ID == Intrinsic::lifetime_end || ID == Intrinsic::assume ||
-      ID == Intrinsic::sideeffect)
+      ID == Intrinsic::sideeffect || ID == Intrinsic::noalias)
     return ID;
   return Intrinsic::not_intrinsic;
 }
 
 /// Find the operand of the GEP that should be checked for consecutive
 /// stores. This ignores trailing indices that have no effect on the final
 /// pointer.
 unsigned llvm::getGEPInductionOperand(const GetElementPtrInst *Gep) {

llvm/lib/Transforms/Vectorize/LoopVectorize.cpp

bool VPRecipeBuilder::tryToWiden(Instruction *I, VPBasicBlock *VPBB,
   };
 
   if (!IsVectorizableOpcode(I->getOpcode()))
     return false;
 
   if (CallInst *CI = dyn_cast<CallInst>(I)) {
     Intrinsic::ID ID = getVectorIntrinsicIDForCall(CI, TLI);
     if (ID && (ID == Intrinsic::assume || ID == Intrinsic::lifetime_end ||
-               ID == Intrinsic::lifetime_start || ID == Intrinsic::sideeffect))
+               ID == Intrinsic::lifetime_start || ID == Intrinsic::sideeffect ||
+               ID == Intrinsic::noalias))
       return false;
   }
 
   auto willWiden = [&](unsigned VF) -> bool {
     if (!isa<PHINode>(I) && (CM.isScalarAfterVectorization(I, VF) ||
                              CM.isProfitableToScalarize(I, VF)))
       return false;
     if (CallInst *CI = dyn_cast<CallInst>(I)) {

llvm/test/Transforms/LoopVectorize/noalias.ll

+; RUN: opt -S -loop-vectorize -force-vector-width=2 -force-vector-interleave=1 < %s | FileCheck %s
+
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
+
+; Make sure we can vectorize loops which contain lifetime markers.
+
+; CHECK-LABEL: @test(
+; CHECK: @llvm.noalias.p0i32
+; CHECK: store <2 x i32>
+
+define void @test(i32 *%d) {
+entry:
+  br label %for.body
+
+for.body:
+  %indvars.iv = phi i64 [ 0, %entry ], [ %indvars.iv.next, %for.body ]
+  %d2 = call i32* @llvm.noalias.p0i32(i32* %d, metadata !1)
+  %arrayidx = getelementptr inbounds i32, i32* %d2, i64 %indvars.iv
+  %v1 = load i32, i32* %arrayidx, align 8
+  store i32 100, i32* %arrayidx, align 8
+  %indvars.iv.next = add i64 %indvars.iv, 1
+  %lftr.wideiv = trunc i64 %indvars.iv.next to i32
+  %exitcond = icmp ne i32 %lftr.wideiv, 128
+  br i1 %exitcond, label %for.body, label %for.end
+
+for.end:
+  ret void
+}
+
+declare i32* @llvm.noalias.p0i32(i32*, metadata) nounwind argmemonly
+
+!0 = !{!0, !"some domain"}
+!1 = !{!1, !0, !"some scope"}
+