JIT: Add a simple Counter dumpable

[jakobbotsch]

This allows easily measuring and outputting a single value at the end of all JIT compilations.

Example use:

@@ -1137,6 +1137,9 @@ void Compiler::optReplaceWidenedIV(unsigned lclNum, unsigned newLclNum, Statemen
     }
 }
 
+static Counter s_widened;
+static DumpOnShutdown d("Widened IVs", &s_widened);
+
 //------------------------------------------------------------------------
 // optInductionVariables: Try and optimize induction variables in the method.
 //
@@ -1275,6 +1278,7 @@ PhaseStatus Compiler::optInductionVariables()
             });
 
             changed |= optSinkWidenedIV(lcl->GetLclNum(), newLclNum, loop);
+            s_widened.Value++;
         }
     }

[ghost]

Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch
See info in area-owners.md if you want to be subscribed.

Issue Details

---

This allows easily measuring and outputting a single value at the end of all JIT compilations.

Example use:

@@ -1137,6 +1137,9 @@ void Compiler::optReplaceWidenedIV(unsigned lclNum, unsigned newLclNum, Statemen
     }
 }
 
+static Counter s_widened;
+static DumpOnShutdown d("Widened IVs", &s_widened);
+
 //------------------------------------------------------------------------
 // optInductionVariables: Try and optimize induction variables in the method.
 //
@@ -1275,6 +1278,7 @@ PhaseStatus Compiler::optInductionVariables()
             });
 
             changed |= optSinkWidenedIV(lcl->GetLclNum(), newLclNum, loop);
+            s_widened.Value++;
         }
     }

Author:	jakobbotsch
Assignees:	jakobbotsch
Labels:	area-CodeGen-coreclr
Milestone:	-

[EgorBo]

oh, didn't know Dumpable existed, I always used just a plain fopen/fprintf 🙂

[ryujit-bot]

Diff results for #98350

Throughput diffs

Throughput diffs for linux/arm64 ran on windows/x64

MinOpts (-0.01% to +0.00%)

Collection	PDIFF
libraries.pmi.linux.arm64.checked.mch	-0.01%

Throughput diffs for osx/arm64 ran on windows/x64

MinOpts (-0.00% to +0.01%)

Collection	PDIFF
realworld.run.osx.arm64.checked.mch	+0.01%

Throughput diffs for windows/arm64 ran on windows/x64

MinOpts (-0.00% to +0.01%)

Collection	PDIFF
libraries.pmi.windows.arm64.checked.mch	+0.01%

Details here

---

[jakobbotsch]

cc @dotnet/jit-contrib PTAL @AndyAyersMS

[jakobbotsch]

oh, didn't know Dumpable existed, I always used just a plain fopen/fprintf 🙂

These convenience types are a bit new, I added them in #92112. I frequently use them to do quick adhoc measurements of stats

[AndyAyersMS]

At some point we ought to reconsider our approach: in-jit aggregation vs external aggregation. I get the appeal of the former (no extra step required!) but I frequently find I prefer the latter (via say JitMetrics=1.

But that's no reason to hold up this PR...