[llvm] [VPlan] Speed up VPSlotTracker by using ModuleSlotTracker (PR #139881)

[via llvm-commits]

llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-vectorizers

@llvm/pr-subscribers-llvm-transforms

Author: Igor Kirillov (igogo-x86)

<details>
<summary>Changes</summary>

Currently, when VPSlotTracker is initialized with a VPlan, its assignName method calls printAsOperand on each underlying instruction. Each such call recomputes slot numbers for the entire function, leading to O(N × M) complexity, where M is the number of instructions in the loop and N is the number of instructions in the function.

This results in slow debug output for large loops. For example, printing costs of all instructions becomes O(M² × N), which is especially painful when enabling verbose dumps.

This patch improves debugging performance by caching slot numbers using ModuleSlotTracker. It avoids redundant recomputation and makes debug output significantly faster.

---
Full diff: https://github.com/llvm/llvm-project/pull/139881.diff


2 Files Affected:

- (modified) llvm/lib/Transforms/Vectorize/VPlan.cpp (+4-1) 
- (modified) llvm/lib/Transforms/Vectorize/VPlanHelpers.h (+13-1) 


``````````diff
diff --git a/llvm/lib/Transforms/Vectorize/VPlan.cpp b/llvm/lib/Transforms/Vectorize/VPlan.cpp
index 167aff737d3fd..a1567fab4eff2 100644
--- a/llvm/lib/Transforms/Vectorize/VPlan.cpp
+++ b/llvm/lib/Transforms/Vectorize/VPlan.cpp
@@ -1441,7 +1441,10 @@ void VPSlotTracker::assignName(const VPValue *V) {
   std::string Name;
   if (UV) {
     raw_string_ostream S(Name);
-    UV->printAsOperand(S, false);
+    if (MST)
+      UV->printAsOperand(S, false, *MST);
+    else
+      UV->printAsOperand(S, false);
   } else
     Name = VPI->getName();
 
diff --git a/llvm/lib/Transforms/Vectorize/VPlanHelpers.h b/llvm/lib/Transforms/Vectorize/VPlanHelpers.h
index 1d42c8f5f3737..0aa0133f4fe1c 100644
--- a/llvm/lib/Transforms/Vectorize/VPlanHelpers.h
+++ b/llvm/lib/Transforms/Vectorize/VPlanHelpers.h
@@ -23,6 +23,7 @@
 #include "llvm/Analysis/DomTreeUpdater.h"
 #include "llvm/Analysis/TargetTransformInfo.h"
 #include "llvm/IR/DebugLoc.h"
+#include "llvm/IR/ModuleSlotTracker.h"
 #include "llvm/Support/InstructionCost.h"
 
 namespace llvm {
@@ -382,14 +383,25 @@ class VPSlotTracker {
   /// Number to assign to the next VPValue without underlying value.
   unsigned NextSlot = 0;
 
+  /// Cache slot indexes to avoid recomputing them on each printAsOperand call.
+  std::unique_ptr<ModuleSlotTracker> MST;
+
   void assignName(const VPValue *V);
   void assignNames(const VPlan &Plan);
   void assignNames(const VPBasicBlock *VPBB);
 
 public:
   VPSlotTracker(const VPlan *Plan = nullptr) {
-    if (Plan)
+    if (Plan) {
+      // This check is required to support unit tests with incomplete IR.
+      if (Function *F =
+              Plan->getScalarHeader()->getIRBasicBlock()->getParent()) {
+        Module *M = F->getParent();
+        MST = std::make_unique<ModuleSlotTracker>(M);
+        MST->incorporateFunction(*F);
+      }
       assignNames(*Plan);
+    }
   }
 
   /// Returns the name assigned to \p V, if there is one, otherwise try to

``````````

</details>


https://github.com/llvm/llvm-project/pull/139881