[PATCH] D59703: Convert a masked.load of a dereferenceable address to an unconditional load

[Philip Reames via Phabricator via llvm-commits]

reames updated this revision to Diff 196105.
reames retitled this revision from "Optimize masked.loads and masked.gathers with a single active lane" to "Convert a masked.load of a dereferenceable address to an unconditional load".
reames edited the summary of this revision.
reames added a comment.

Split the patch.  This review is now only the masked.load part.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D59703/new/

https://reviews.llvm.org/D59703

Files:
  lib/Transforms/InstCombine/InstCombineCalls.cpp
  test/Transforms/InstCombine/masked_intrinsics.ll


Index: test/Transforms/InstCombine/masked_intrinsics.ll
===================================================================
--- test/Transforms/InstCombine/masked_intrinsics.ll
+++ test/Transforms/InstCombine/masked_intrinsics.ll
@@ -87,8 +87,9 @@
 ; CHECK-LABEL: @load_speculative(
 ; CHECK-NEXT:    [[PTV1:%.*]] = insertelement <2 x double> undef, double [[PT:%.*]], i64 0
 ; CHECK-NEXT:    [[PTV2:%.*]] = shufflevector <2 x double> [[PTV1]], <2 x double> undef, <2 x i32> zeroinitializer
-; CHECK-NEXT:    [[RES:%.*]] = call <2 x double> @llvm.masked.load.v2f64.p0v2f64(<2 x double>* nonnull [[PTR:%.*]], i32 4, <2 x i1> [[MASK:%.*]], <2 x double> [[PTV2]])
-; CHECK-NEXT:    ret <2 x double> [[RES]]
+; CHECK-NEXT:    [[UNMASKEDLOAD:%.*]] = load <2 x double>, <2 x double>* [[PTR:%.*]], align 4
+; CHECK-NEXT:    [[TMP1:%.*]] = select <2 x i1> [[MASK:%.*]], <2 x double> [[UNMASKEDLOAD]], <2 x double> [[PTV2]]
+; CHECK-NEXT:    ret <2 x double> [[TMP1]]
 ;
   double %pt, <2 x i1> %mask)  {
   %ptv1 = insertelement <2 x double> undef, double %pt, i64 0
Index: lib/Transforms/InstCombine/InstCombineCalls.cpp
===================================================================
--- lib/Transforms/InstCombine/InstCombineCalls.cpp
+++ lib/Transforms/InstCombine/InstCombineCalls.cpp
@@ -21,6 +21,7 @@
 #include "llvm/ADT/Statistic.h"
 #include "llvm/ADT/Twine.h"
 #include "llvm/Analysis/AssumptionCache.h"
+#include "llvm/Analysis/Loads.h"
 #include "llvm/Analysis/InstructionSimplify.h"
 #include "llvm/Analysis/MemoryBuiltins.h"
 #include "llvm/Transforms/Utils/Local.h"
@@ -1165,19 +1166,29 @@
 }
 
 // TODO, Obvious Missing Transforms:
-// * Dereferenceable address -> speculative load/select
+// * Single active lane to scalar masked load
 // * Narrow width by halfs excluding zero/undef lanes
 static Value *simplifyMaskedLoad(const IntrinsicInst &II,
                                  InstCombiner::BuilderTy &Builder) {
+  Value *LoadPtr = II.getArgOperand(0);
+  unsigned Alignment = cast<ConstantInt>(II.getArgOperand(1))->getZExtValue();
+
   // If the mask is all ones or undefs, this is a plain vector load of the 1st
   // argument.
-  if (maskIsAllOneOrUndef(II.getArgOperand(2))) {
-    Value *LoadPtr = II.getArgOperand(0);
-    unsigned Alignment = cast<ConstantInt>(II.getArgOperand(1))->getZExtValue();
+  if (maskIsAllOneOrUndef(II.getArgOperand(2)))
     return Builder.CreateAlignedLoad(II.getType(), LoadPtr, Alignment,
                                      "unmaskedload");
-  }
 
+  // If we can unconditionally load from this address, replace with a
+  // load/select idiom. TODO: use DT for context sensative query
+  if (isDereferenceableAndAlignedPointer(LoadPtr, Alignment,
+                                         II.getModule()->getDataLayout(),
+                                         &II, nullptr)) {
+    auto *LI = Builder.CreateAlignedLoad(II.getType(), LoadPtr, Alignment,
+                                         "unmaskedload");
+    return Builder.CreateSelect(II.getArgOperand(2), LI, II.getArgOperand(3));
+  }
+  
   return nullptr;
 }
 


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D59703.196105.patch
Type: text/x-patch
Size: 3091 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20190422/7321933b/attachment.bin>