[llvm] [IPT] Remove ipt.NumInstScanned statistic (PR #168515)
via llvm-commits
llvm-commits at lists.llvm.org
Tue Nov 18 03:21:48 PST 2025
llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-llvm-analysis
Author: Luke Lau (lukel97)
<details>
<summary>Changes</summary>
The NumInstScanned statistic is non-determinstic across multiple identical invocations of LLVM, and leads to noise when trying to diff LLVM statistics with e.g. ./utils/tdiff.py in llvm-test-suite.
My understanding is that it's non-deterministic because the users of IPT's hasSpecialInstructions/isPreceededBySpecialInstruction API aren't deterministic themselves.
This PR removes it and fixes #<!-- -->157598. This is just a small quality-of-life improvement for the ./utils/tdiff.py workflow, but happy to leave the statistic in if others are using it.
---
Full diff: https://github.com/llvm/llvm-project/pull/168515.diff
1 Files Affected:
- (modified) llvm/lib/Analysis/InstructionPrecedenceTracking.cpp (-5)
``````````diff
diff --git a/llvm/lib/Analysis/InstructionPrecedenceTracking.cpp b/llvm/lib/Analysis/InstructionPrecedenceTracking.cpp
index a2a9c94c14ae8..8bebd465ff7d3 100644
--- a/llvm/lib/Analysis/InstructionPrecedenceTracking.cpp
+++ b/llvm/lib/Analysis/InstructionPrecedenceTracking.cpp
@@ -19,15 +19,11 @@
#include "llvm/Analysis/InstructionPrecedenceTracking.h"
#include "llvm/Analysis/ValueTracking.h"
-#include "llvm/ADT/Statistic.h"
#include "llvm/IR/PatternMatch.h"
#include "llvm/Support/CommandLine.h"
using namespace llvm;
-#define DEBUG_TYPE "ipt"
-STATISTIC(NumInstScanned, "Number of insts scanned while updating ibt");
-
#ifndef NDEBUG
static cl::opt<bool> ExpensiveAsserts(
"ipt-expensive-asserts",
@@ -50,7 +46,6 @@ const Instruction *InstructionPrecedenceTracking::getFirstSpecialInstruction(
auto [It, Inserted] = FirstSpecialInsts.try_emplace(BB);
if (Inserted) {
for (const auto &I : *BB) {
- NumInstScanned++;
if (isSpecialInstruction(&I)) {
It->second = &I;
break;
``````````
</details>
https://github.com/llvm/llvm-project/pull/168515
More information about the llvm-commits
mailing list