diff --git a/llvm/lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp b/llvm/lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp
index 907f82ed7fc52..fec0ebcd70593 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPUPromoteAlloca.cpp
@@ -1556,12 +1556,34 @@ bool AMDGPUPromoteAllocaImpl::tryPromoteAllocaToLDS(AllocaInst &I,
     case Intrinsic::invariant_start:
     case Intrinsic::invariant_end:
     case Intrinsic::launder_invariant_group:
-    case Intrinsic::strip_invariant_group:
+    case Intrinsic::strip_invariant_group: {
+      SmallVector<Type *> ArgTy;
+      SmallVector<Value *> Args;
+      if (Intr->getIntrinsicID() == Intrinsic::invariant_start) {
+        Value *Size = Intr->getArgOperand(0);
+        ArgTy.emplace_back(Offset->getType());
+        Args.emplace_back(Size);
+        Args.emplace_back(Offset);
+      } else if (Intr->getIntrinsicID() == Intrinsic::invariant_end) {
+        Value *InvariantPtr = Intr->getArgOperand(0);
+        Value *Size = Intr->getArgOperand(1);
+        ArgTy.emplace_back(Offset->getType());
+        Args.emplace_back(InvariantPtr);
+        Args.emplace_back(Size);
+        Args.emplace_back(Offset);
+      } else {
+        ArgTy.emplace_back(Offset->getType());
+        Args.emplace_back(Offset);
+      }
+      Function *F = Intrinsic::getOrInsertDeclaration(
+          Intr->getModule(), Intr->getIntrinsicID(), ArgTy);
+      CallInst *NewIntr =
+          CallInst::Create(F, Args, Intr->getName(), Intr->getIterator());
+      Intr->mutateType(NewIntr->getType());
+      Intr->replaceAllUsesWith(NewIntr);
       Intr->eraseFromParent();
-      // FIXME: I think the invariant marker should still theoretically apply,
-      // but the intrinsics need to be changed to accept pointers with any
-      // address space.
       continue;
+    }
     case Intrinsic::objectsize: {
       Value *Src = Intr->getOperand(0);
 
diff --git a/llvm/test/CodeGen/AMDGPU/promote-alloca-invariant-marks.ll b/llvm/test/CodeGen/AMDGPU/promote-alloca-invariant-marks.ll
new file mode 100644
index 0000000000000..fca4be5e76dae
--- /dev/null
+++ b/llvm/test/CodeGen/AMDGPU/promote-alloca-invariant-marks.ll
@@ -0,0 +1,75 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes=amdgpu-promote-alloca %s -o - | FileCheck %s
+
+declare ptr @llvm.invariant.start.p5(i64, ptr addrspace(5) nocapture)
+declare void @llvm.invariant.end.p5(ptr, i64, ptr addrspace(5) nocapture)
+declare ptr addrspace(5) @llvm.launder.invariant.group.p5(ptr addrspace(5))
+declare ptr addrspace(5) @llvm.strip.invariant.group.p5(ptr addrspace(5))
+
+define amdgpu_kernel void @use_invariant_start_and_end() {
+; CHECK-LABEL: define amdgpu_kernel void @use_invariant_start_and_end() {
+; CHECK-NEXT:  [[BB:.*:]]
+; CHECK-NEXT:    [[TMP0:%.*]] = call noalias nonnull dereferenceable(64) ptr addrspace(4) @llvm.amdgcn.dispatch.ptr()
+; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr addrspace(4) [[TMP0]], i64 1
+; CHECK-NEXT:    [[TMP2:%.*]] = load i32, ptr addrspace(4) [[TMP1]], align 4, !invariant.load [[META0:![0-9]+]]
+; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i32, ptr addrspace(4) [[TMP0]], i64 2
+; CHECK-NEXT:    [[TMP4:%.*]] = load i32, ptr addrspace(4) [[TMP3]], align 4, !range [[RNG1:![0-9]+]], !invariant.load [[META0]]
+; CHECK-NEXT:    [[TMP5:%.*]] = lshr i32 [[TMP2]], 16
+; CHECK-NEXT:    [[TMP6:%.*]] = call range(i32 0, 1024) i32 @llvm.amdgcn.workitem.id.x()
+; CHECK-NEXT:    [[TMP7:%.*]] = call range(i32 0, 1024) i32 @llvm.amdgcn.workitem.id.y()
+; CHECK-NEXT:    [[TMP8:%.*]] = call range(i32 0, 1024) i32 @llvm.amdgcn.workitem.id.z()
+; CHECK-NEXT:    [[TMP9:%.*]] = mul nuw nsw i32 [[TMP5]], [[TMP4]]
+; CHECK-NEXT:    [[TMP10:%.*]] = mul i32 [[TMP9]], [[TMP6]]
+; CHECK-NEXT:    [[TMP11:%.*]] = mul nuw nsw i32 [[TMP7]], [[TMP4]]
+; CHECK-NEXT:    [[TMP12:%.*]] = add i32 [[TMP10]], [[TMP11]]
+; CHECK-NEXT:    [[TMP13:%.*]] = add i32 [[TMP12]], [[TMP8]]
+; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [1024 x i32], ptr addrspace(3) @use_invariant_start_and_end.alloca, i32 0, i32 [[TMP13]]
+; CHECK-NEXT:    [[INVARIENT1:%.*]] = call ptr @llvm.invariant.start.p3(i64 0, ptr addrspace(3) [[TMP14]])
+; CHECK-NEXT:    store <2 x i1> zeroinitializer, ptr [[INVARIENT1]], align 1
+; CHECK-NEXT:    call void @llvm.invariant.end.p3(ptr [[INVARIENT1]], i64 0, ptr addrspace(3) [[TMP14]])
+; CHECK-NEXT:    ret void
+;
+bb:
+  %alloca = alloca i32, align 4, addrspace(5)
+  %invarient = call ptr @llvm.invariant.start.p5(i64 0, ptr addrspace(5) %alloca)
+  store <2 x i1> zeroinitializer, ptr %invarient, align 1
+  call void @llvm.invariant.end.p5(ptr %invarient, i64 0, ptr addrspace(5) %alloca)
+  ret void
+}
+
+define amdgpu_kernel void @use_invariant_group_and_strip() {
+; CHECK-LABEL: define amdgpu_kernel void @use_invariant_group_and_strip() {
+; CHECK-NEXT:  [[BB:.*:]]
+; CHECK-NEXT:    [[TMP0:%.*]] = call noalias nonnull dereferenceable(64) ptr addrspace(4) @llvm.amdgcn.dispatch.ptr()
+; CHECK-NEXT:    [[TMP1:%.*]] = getelementptr inbounds i32, ptr addrspace(4) [[TMP0]], i64 1
+; CHECK-NEXT:    [[TMP2:%.*]] = load i32, ptr addrspace(4) [[TMP1]], align 4, !invariant.load [[META0]]
+; CHECK-NEXT:    [[TMP3:%.*]] = getelementptr inbounds i32, ptr addrspace(4) [[TMP0]], i64 2
+; CHECK-NEXT:    [[TMP4:%.*]] = load i32, ptr addrspace(4) [[TMP3]], align 4, !range [[RNG1]], !invariant.load [[META0]]
+; CHECK-NEXT:    [[TMP5:%.*]] = lshr i32 [[TMP2]], 16
+; CHECK-NEXT:    [[TMP6:%.*]] = call range(i32 0, 1024) i32 @llvm.amdgcn.workitem.id.x()
+; CHECK-NEXT:    [[TMP7:%.*]] = call range(i32 0, 1024) i32 @llvm.amdgcn.workitem.id.y()
+; CHECK-NEXT:    [[TMP8:%.*]] = call range(i32 0, 1024) i32 @llvm.amdgcn.workitem.id.z()
+; CHECK-NEXT:    [[TMP9:%.*]] = mul nuw nsw i32 [[TMP5]], [[TMP4]]
+; CHECK-NEXT:    [[TMP10:%.*]] = mul i32 [[TMP9]], [[TMP6]]
+; CHECK-NEXT:    [[TMP11:%.*]] = mul nuw nsw i32 [[TMP7]], [[TMP4]]
+; CHECK-NEXT:    [[TMP12:%.*]] = add i32 [[TMP10]], [[TMP11]]
+; CHECK-NEXT:    [[TMP13:%.*]] = add i32 [[TMP12]], [[TMP8]]
+; CHECK-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [1024 x i32], ptr addrspace(3) @use_invariant_group_and_strip.alloca, i32 0, i32 [[TMP13]]
+; CHECK-NEXT:    [[INVARIENT2:%.*]] = call ptr addrspace(3) @llvm.launder.invariant.group.p3(ptr addrspace(3) [[TMP14]])
+; CHECK-NEXT:    store <2 x i1> zeroinitializer, ptr addrspace(3) [[INVARIENT2]], align 1
+; CHECK-NEXT:    [[STRIP1:%.*]] = call ptr addrspace(3) @llvm.strip.invariant.group.p3(ptr addrspace(3) [[TMP14]])
+; CHECK-NEXT:    store <2 x i1> zeroinitializer, ptr addrspace(3) [[STRIP1]], align 1
+; CHECK-NEXT:    ret void
+;
+bb:
+  %alloca = alloca i32, align 4, addrspace(5)
+  %invarient = call ptr addrspace(5) @llvm.launder.invariant.group.p5(ptr addrspace(5) %alloca)
+  store <2 x i1> zeroinitializer, ptr addrspace(5) %invarient, align 1
+  %strip = call ptr addrspace(5) @llvm.strip.invariant.group.p5(ptr addrspace(5) %alloca)
+  store <2 x i1> zeroinitializer, ptr addrspace(5) %strip, align 1
+  ret void
+}
+;.
+; CHECK: [[META0]] = !{}
+; CHECK: [[RNG1]] = !{i32 0, i32 1025}
+;.