commit dbbac8f812a866b1b53f3007721f66038d208549
author Aart Bik <ajcbik@google.com> Fri Sep 01 13:06:08 2017 -0700
committer Aart Bik <ajcbik@google.com> Thu Sep 21 10:20:55 2017 -0700
tree 05cecd927afccd33fc1c14b39ada47e86873f560
parent 2406bf17998e15bd40677a907beb3e9c41facce4

Implement Sum-of-Abs-Differences idiom recognition.

Rationale:
Currently just on ARM64 (x86 lacks proper support),
using the SAD idiom yields great speedup on loops
that compute the sum-of-abs-difference operation.
Also includes some refinements around type conversions.

Speedup ExoPlayerAudio (golem run):
1.3x on ARM64
1.1x on x86

Test: test-art-host test-art-target

Bug: 64091002

Change-Id: Ia2b711d2bc23609a2ed50493dfe6719eedfe0130

compiler/optimizing/code_generator_vector_x86.cc

diff --git a/compiler/optimizing/code_generator_vector_x86.cc b/compiler/optimizing/code_generator_vector_x86.cc
index 37190f8..5a012e7 100644
--- a/compiler/optimizing/code_generator_vector_x86.cc
+++ b/compiler/optimizing/code_generator_vector_x86.cc
@@ -51,7 +51,6 @@
                                     : Location::RequiresFpuRegister());
       locations->SetOut(is_zero ? Location::RequiresFpuRegister()
                                 : Location::SameAsFirstInput());
-
       break;
     default:
       LOG(FATAL) << "Unsupported SIMD type";
@@ -1033,12 +1032,42 @@
   }
 }
 
-void LocationsBuilderX86::VisitVecMultiplyAccumulate(HVecMultiplyAccumulate* instr) {
-  LOG(FATAL) << "No SIMD for " << instr->GetId();
+// Helper to set up locations for vector accumulations.
+static void CreateVecAccumLocations(ArenaAllocator* arena, HVecOperation* instruction) {
+  LocationSummary* locations = new (arena) LocationSummary(instruction);
+  switch (instruction->GetPackedType()) {
+    case Primitive::kPrimByte:
+    case Primitive::kPrimChar:
+    case Primitive::kPrimShort:
+    case Primitive::kPrimInt:
+    case Primitive::kPrimLong:
+      locations->SetInAt(0, Location::RequiresFpuRegister());
+      locations->SetInAt(1, Location::RequiresFpuRegister());
+      locations->SetInAt(2, Location::RequiresFpuRegister());
+      locations->SetOut(Location::SameAsFirstInput());
+      break;
+    default:
+      LOG(FATAL) << "Unsupported SIMD type";
+      UNREACHABLE();
+  }
 }
 
-void InstructionCodeGeneratorX86::VisitVecMultiplyAccumulate(HVecMultiplyAccumulate* instr) {
-  LOG(FATAL) << "No SIMD for " << instr->GetId();
+void LocationsBuilderX86::VisitVecMultiplyAccumulate(HVecMultiplyAccumulate* instruction) {
+  CreateVecAccumLocations(GetGraph()->GetArena(), instruction);
+}
+
+void InstructionCodeGeneratorX86::VisitVecMultiplyAccumulate(HVecMultiplyAccumulate* instruction) {
+  // TODO: pmaddwd?
+  LOG(FATAL) << "No SIMD for " << instruction->GetId();
+}
+
+void LocationsBuilderX86::VisitVecSADAccumulate(HVecSADAccumulate* instruction) {
+  CreateVecAccumLocations(GetGraph()->GetArena(), instruction);
+}
+
+void InstructionCodeGeneratorX86::VisitVecSADAccumulate(HVecSADAccumulate* instruction) {
+  // TODO: psadbw for unsigned?
+  LOG(FATAL) << "No SIMD for " << instruction->GetId();
 }
 
 // Helper to set up locations for vector memory operations.