[AMDGPU] gfx1010 DS implementation
Differential Revision: https://reviews.llvm.org/D61332
llvm-svn: 359696
diff --git a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
index e226914..6b82ed4 100644
--- a/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
+++ b/llvm/lib/Target/AMDGPU/SIISelLowering.cpp
@@ -6649,6 +6649,11 @@
std::tie(Ops[0], Ops[1]) = expandUnalignedLoad(Load, DAG);
return DAG.getMergeValues(Ops, DL);
}
+ if (Subtarget->hasLDSMisalignedBug() &&
+ AS == AMDGPUAS::FLAT_ADDRESS &&
+ Alignment < MemVT.getStoreSize() && MemVT.getSizeInBits() > 32) {
+ return SplitVectorLoad(Op, DAG);
+ }
MachineFunction &MF = DAG.getMachineFunction();
SIMachineFunctionInfo *MFI = MF.getInfo<SIMachineFunctionInfo>();
@@ -7110,6 +7115,12 @@
return expandUnalignedStore(Store, DAG);
}
+ if (Subtarget->hasLDSMisalignedBug() &&
+ AS == AMDGPUAS::FLAT_ADDRESS &&
+ Store->getAlignment() < VT.getStoreSize() && VT.getSizeInBits() > 32) {
+ return SplitVectorStore(Op, DAG);
+ }
+
MachineFunction &MF = DAG.getMachineFunction();
SIMachineFunctionInfo *MFI = MF.getInfo<SIMachineFunctionInfo>();
// If there is a possibilty that flat instruction access scratch memory