diff --git a/llvm/lib/Target/PowerPC/PPCRegisterInfo.cpp b/llvm/lib/Target/PowerPC/PPCRegisterInfo.cpp
--- a/llvm/lib/Target/PowerPC/PPCRegisterInfo.cpp
+++ b/llvm/lib/Target/PowerPC/PPCRegisterInfo.cpp
@@ -1491,6 +1491,18 @@
       Offset += MFI.getStackSize();
   }
 
+  // If we encounter an LQ/STQ instruction with an offset that fits
+  // in the instruction, then generate the instruction with the correct
+  // offset and base address.
+  if (((OpC == PPC::LQ && isInt<12>(Offset)) ||
+       (OpC == PPC::STQ && isInt<16>(Offset))) &&
+      ((Offset % offsetMinAlign(MI)) == 0)) {
+    Register StackReg = MI.getOperand(FIOperandNum).getReg();
+    MI.getOperand(1).ChangeToImmediate(Offset);
+    MI.getOperand(2).ChangeToRegister(StackReg, false);
+    return;
+  }
+
   // If we encounter an LXVP/STXVP with an offset that doesn't fit, we can
   // transform it to the prefixed version so we don't have to use the XForm.
   if ((OpC == PPC::LXVP || OpC == PPC::STXVP) &&
diff --git a/llvm/test/CodeGen/PowerPC/load-store-quadword.ll b/llvm/test/CodeGen/PowerPC/load-store-quadword.ll
new file mode 100644
--- /dev/null
+++ b/llvm/test/CodeGen/PowerPC/load-store-quadword.ll
@@ -0,0 +1,30 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mcpu=pwr8 -verify-machineinstrs -ppc-asm-full-reg-names -o - %s | FileCheck %s
+
+target triple = "powerpc64le-unknown-linux-gnu"
+
+%struct.StructA = type { [16 x i8] }
+
+@s1 = dso_local global %struct.StructA { [16 x i8] c"\0B\0C\0D\0E\0F\10\11\12\13\14\15\16\17\18\19\1A" }, align 16
+
+; Function Attrs: noinline nounwind optnone uwtable
+define dso_local void @test() {
+; CHECK-LABEL: test:
+; CHECK:       # %bb.0: # %entry
+; CHECK:    stq r4, -16(r1)
+; CHECK-NEXT:    sync
+; CHECK-NEXT:    lq r4, -16(r1)
+entry:
+  %s2 = alloca %struct.StructA, align 16
+  %s3 = alloca %struct.StructA, align 16
+  %agg.tmp.ensured = alloca %struct.StructA, align 16
+  call void @llvm.memcpy.p0.p0.i64(ptr align 16 %agg.tmp.ensured, ptr align 16 @s1, i64 16, i1 false)
+  %0 = load i128, ptr %agg.tmp.ensured, align 16
+  store atomic i128 %0, ptr %s2 seq_cst, align 16
+  %atomic-load = load atomic i128, ptr %s2 seq_cst, align 16
+  store i128 %atomic-load, ptr %s3, align 16
+  ret void
+}
+
+; Function Attrs: argmemonly nofree nounwind willreturn
+declare void @llvm.memcpy.p0.p0.i64(ptr noalias nocapture writeonly, ptr noalias nocapture readonly, i64, i1 immarg)