[PATCH] D120780: speeding up ext-tsp for huge instances

Sergey Pupyrev via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Tue Mar 1 15:16:49 PST 2022


spupyrev updated this revision to Diff 412274.
spupyrev added a comment.

minor


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D120780/new/

https://reviews.llvm.org/D120780

Files:
  llvm/lib/CodeGen/MachineBlockPlacement.cpp
  llvm/lib/Transforms/Utils/CodeLayout.cpp


Index: llvm/lib/Transforms/Utils/CodeLayout.cpp
===================================================================
--- llvm/lib/Transforms/Utils/CodeLayout.cpp
+++ llvm/lib/Transforms/Utils/CodeLayout.cpp
@@ -49,6 +49,11 @@
     cl::desc("Enable machine block placement based on the ext-tsp model, "
              "optimizing I-cache utilization."));
 
+cl::opt<bool> ApplyExtTspApplyNoProfile(
+    "ext-tsp-apply-no-profile",
+    cl::desc("Whether to apply ext-tsp placement for instances w/o profile"),
+    cl::init(true), cl::Hidden, cl::ZeroOrMore);
+
 // Algorithm-specific constants. The values are tuned for the best performance
 // of large-scale front-end bound binaries.
 static cl::opt<double>
@@ -67,6 +72,12 @@
     "ext-tsp-backward-distance", cl::Hidden, cl::init(640),
     cl::desc("The maximum distance (in bytes) of a backward jump for ExtTSP"));
 
+// The maximum size of a chain created by the algorithm. The size is bounded
+// so that the algorithm can efficiently process extremely large instance.
+static cl::opt<unsigned>
+    MaxChainSize("ext-tsp-max-chain-size", cl::Hidden, cl::init(4096),
+                 cl::desc("The maximum size of a chain to create."));
+
 // The maximum size of a chain for splitting. Larger values of the threshold
 // may yield better quality at the cost of worsen run-time.
 static cl::opt<unsigned> ChainSplitThreshold(
@@ -226,6 +237,8 @@
 
   const std::vector<Block *> &blocks() const { return Blocks; }
 
+  size_t numBlocks() const { return Blocks.size(); }
+
   const std::vector<std::pair<Chain *, ChainEdge *>> &edges() const {
     return Edges;
   }
@@ -502,7 +515,7 @@
     AllEdges.reserve(AllJumps.size());
     for (auto &Block : AllBlocks) {
       for (auto &Jump : Block.OutJumps) {
-        const auto SuccBlock = Jump->Target;
+        auto SuccBlock = Jump->Target;
         auto CurEdge = Block.CurChain->getEdge(SuccBlock->CurChain);
         // this edge is already present in the graph
         if (CurEdge != nullptr) {
@@ -592,6 +605,10 @@
           if (ChainPred == ChainSucc)
             continue;
 
+          // Stop early if the combined chain violates the maximum allowed size
+          if (ChainPred->numBlocks() + ChainSucc->numBlocks() >= MaxChainSize)
+            continue;
+
           // Compute the gain of merging the two chains
           auto CurGain = getBestMergeGain(ChainPred, ChainSucc, ChainEdge);
           if (CurGain.score() <= EPS)
Index: llvm/lib/CodeGen/MachineBlockPlacement.cpp
===================================================================
--- llvm/lib/CodeGen/MachineBlockPlacement.cpp
+++ llvm/lib/CodeGen/MachineBlockPlacement.cpp
@@ -201,6 +201,7 @@
     cl::Hidden);
 
 extern cl::opt<bool> EnableExtTspBlockPlacement;
+extern cl::opt<bool> ApplyExtTspApplyNoProfile;
 
 namespace llvm {
 extern cl::opt<unsigned> StaticLikelyProb;
@@ -3419,7 +3420,8 @@
   }
 
   // Apply a post-processing optimizing block placement.
-  if (MF.size() >= 3 && EnableExtTspBlockPlacement) {
+  if (MF.size() >= 3 && EnableExtTspBlockPlacement &&
+      (ApplyExtTspApplyNoProfile || MF.getFunction().hasProfileData())) {
     // Find a new placement and modify the layout of the blocks in the function.
     applyExtTsp();
 


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D120780.412274.patch
Type: text/x-patch
Size: 3243 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20220301/a8fea48a/attachment.bin>


More information about the llvm-commits mailing list