[PATCH] D139448: [llvm-exegesis] Dry run mode

Clement Courbet via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Wed Dec 7 06:38:35 PST 2022


courbet added inline comments.


================
Comment at: llvm/tools/llvm-exegesis/lib/BenchmarkRunner.h:39
+                           InstructionBenchmark::ModeE Mode,
+                           bool BenchmarkDryRun);
 
----------------
lebedev.ri wrote:
> courbet wrote:
> > Instead of plumbing `BenchmarkDryRun` inside this class, why can't we just not create a `BenchmarkRunner` at all in `llvm-exegesis.cpp` when the flag is true ?
> Because that, by definition, does not do what i want it to do.
> I want `BenchmarkRunner::runConfiguration()` to get called
> up to and including actually assembling the final code that we'd measure
> in `for (const std::unique_ptr<const SnippetRepetitor> &Repetitor : Repetitors) {`.
Right, I read too fast. Then le'ts rename `BenchmarkDryRun` to `PerformMeasurements`, to avoid the terminology conflict between the "Run" in "BenchmarkRunner" and "Run" in "BenchmarkDryRun". In the former, "Run" means "generate snippet and measure", in the latter it means "measure".


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D139448/new/

https://reviews.llvm.org/D139448



More information about the llvm-commits mailing list