[PATCH] D14603: [X86ISelLowering] Add additional support for multiplication-to-shift conversion.

[Chen Li via llvm-commits]

chenli created this revision.
chenli added a reviewer: craig.topper.
chenli added a subscriber: llvm-commits.
Herald added a subscriber: aemerson.

This patch adds support of conversion (mul x, 2^N + 1) => (add (shl x, N), x) and (mul x, 2^N - 1) => (sub (shl x, N), x) if the multiplication can not be converted to LEA + SHL or LEA + LEA. LLVM has already supported this on ARM, and it should also be useful on X86. Note the patch currently only applies to cases where the constant operand is positive, and I am planing to add another patch to support negative cases after this.

http://reviews.llvm.org/D14603

Files:
  lib/Target/X86/X86ISelLowering.cpp
  test/CodeGen/X86/imul.ll

Index: test/CodeGen/X86/imul.ll
===================================================================
--- test/CodeGen/X86/imul.ll
+++ test/CodeGen/X86/imul.ll
@@ -126,3 +126,25 @@
     %mul = mul i32 %A, 40
     ret i32 %mul
 }
+
+define i32 @mul33_32(i32 %A) {
+; X64-LABEL: mul33_32:
+; X64: shll
+; X64-NEXT: leal
+; X86-LABEL: mul33_32:
+; X86: shll
+; X86-NEXT: addl
+    %mul = mul i32 %A, 33
+    ret i32 %mul
+}
+
+define i32 @mul31_32(i32 %A) {
+; X64-LABEL: mul31_32:
+; X64: shll
+; X64-NEXT: subl
+; X86-LABEL: mul31_32:
+; X86: shll
+; X86-NEXT: subl
+    %mul = mul i32 %A, 31
+    ret i32 %mul
+}
\ No newline at end of file
Index: lib/Target/X86/X86ISelLowering.cpp
===================================================================
--- lib/Target/X86/X86ISelLowering.cpp
+++ lib/Target/X86/X86ISelLowering.cpp
@@ -24274,18 +24274,19 @@
     MulAmt1 = 3;
     MulAmt2 = MulAmt / 3;
   }
+
+  SDLoc DL(N);
+  SDValue NewMul;
   if (MulAmt2 &&
       (isPowerOf2_64(MulAmt2) || MulAmt2 == 3 || MulAmt2 == 5 || MulAmt2 == 9)){
-    SDLoc DL(N);
 
     if (isPowerOf2_64(MulAmt2) &&
         !(N->hasOneUse() && N->use_begin()->getOpcode() == ISD::ADD))
       // If second multiplifer is pow2, issue it first. We want the multiply by
       // 3, 5, or 9 to be folded into the addressing mode unless the lone use
       // is an add.
       std::swap(MulAmt1, MulAmt2);
 
-    SDValue NewMul;
     if (isPowerOf2_64(MulAmt1))
       NewMul = DAG.getNode(ISD::SHL, DL, VT, N->getOperand(0),
                            DAG.getConstant(Log2_64(MulAmt1), DL, MVT::i8));
@@ -24299,10 +24300,28 @@
     else
       NewMul = DAG.getNode(X86ISD::MUL_IMM, DL, VT, NewMul,
                            DAG.getConstant(MulAmt2, DL, VT));
+  }
 
+  if (!NewMul) {
+    if (isPowerOf2_64(MulAmt - 1))
+      // (mul x, 2^N + 1) => (add (shl x, N), x)
+      NewMul = DAG.getNode(ISD::ADD, DL, VT, N->getOperand(0),
+                                DAG.getNode(ISD::SHL, DL, VT, N->getOperand(0),
+                                DAG.getConstant(Log2_64(MulAmt - 1), DL,
+                                MVT::i8)));
+
+    else if (isPowerOf2_64(MulAmt + 1))
+      // (mul x, 2^N - 1) => (sub (shl x, N), x)
+      NewMul = DAG.getNode(ISD::SUB, DL, VT, DAG.getNode(ISD::SHL, DL, VT,
+                                N->getOperand(0),
+                                DAG.getConstant(Log2_64(MulAmt + 1),
+                                DL, MVT::i8)), N->getOperand(0));
+  }
+
+  if (NewMul)
     // Do not add new nodes to DAG combiner worklist.
     DCI.CombineTo(N, NewMul, false);
-  }
+
   return SDValue();
 }
 


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D14603.40008.patch
Type: text/x-patch
Size: 2620 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20151112/7aa199c3/attachment.bin>