[PATCH] D63700: [X86] Don't a vzext_movl in LowerBuildVectorv16i8/LowerBuildVectorv8i16 if there are no zeroes in the vector we're building.

[Craig Topper via Phabricator via llvm-commits]

craig.topper created this revision.
craig.topper added reviewers: spatel, RKSimon.
Herald added a subscriber: hiraditya.
Herald added a project: LLVM.

In LowerBuildVectorv16i8 we took care to use an any_extend if the first pair is in the lower 16-bits of the vector and no elements are 0. So bits [31:16] will be undefined. But we still emitted a vzext_movl to ensure that bits [127:32] are 0. If we don't need any zeroes we should be consistent and make all of 127:16 undefined.

In LowerBuildVectorv8i16 we can just delete the vzext_movl code because we only use the scalar_to_vector when there are no zeroes. So the vzext_movl is always unnecessary.

Found while investigating whether (vzext_movl (scalar_to_vector (loadi32)) patterns are necessary. At least one of the cases where they were necessary was where the loadi32 matched 32-bit aligned 16-bit extload. Seemed weird that we required vzext_movl for that case.


https://reviews.llvm.org/D63700

Files:
  llvm/lib/Target/X86/X86ISelLowering.cpp


Index: llvm/lib/Target/X86/X86ISelLowering.cpp
===================================================================
--- llvm/lib/Target/X86/X86ISelLowering.cpp
+++ llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -7112,7 +7112,7 @@
 
     // If the build vector contains zeros or our first insertion is not the
     // first index then insert into zero vector to break any register
-    // dependency else use SCALAR_TO_VECTOR/VZEXT_MOVL.
+    // dependency else use SCALAR_TO_VECTOR.
     if (First) {
       First = false;
       if (NumZero || 0 != i)
@@ -7121,7 +7121,6 @@
         assert(0 == i && "Expected insertion into zero-index");
         V = DAG.getAnyExtOrTrunc(Op.getOperand(i), dl, MVT::i32);
         V = DAG.getNode(ISD::SCALAR_TO_VECTOR, dl, MVT::v4i32, V);
-        V = DAG.getNode(X86ISD::VZEXT_MOVL, dl, MVT::v4i32, V);
         V = DAG.getBitcast(VT, V);
         continue;
       }
@@ -7184,7 +7183,8 @@
           V = NumZero ? DAG.getZExtOrTrunc(ThisElt, dl, MVT::i32)
                       : DAG.getAnyExtOrTrunc(ThisElt, dl, MVT::i32);
           V = DAG.getNode(ISD::SCALAR_TO_VECTOR, dl, MVT::v4i32, V);
-          V = DAG.getNode(X86ISD::VZEXT_MOVL, dl, MVT::v4i32, V);
+          if (NumZero)
+            V = DAG.getNode(X86ISD::VZEXT_MOVL, dl, MVT::v4i32, V);
           V = DAG.getBitcast(MVT::v8i16, V);
         } else {
           V = DAG.getNode(ISD::INSERT_VECTOR_ELT, dl, MVT::v8i16, V, ThisElt,


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D63700.206155.patch
Type: text/x-patch
Size: 1431 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20190624/a912fb0c/attachment.bin>