Skip to content

[NFC][IRCE] Don't require LoopStructure to determine IRCE profitability #116384

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged

Conversation

JanJecmen
Copy link
Contributor

This refactoring hoists the profitability check earlier in the pipeline, so that for loops that are not profitable to transform there is no iteration over the basic blocks or LoopStructure computation.

Motivated by PR #104659 that tweaks how the profitability of individual branches is evaluated.

This refactoring hoists the profitability check earlier in the pipeline, so that for loops that are not profitable to transform there is no iteration over the basic blocks or LoopStructure computation.

Motivated by PR llvm#104659 that tweaks how the profitability of individual branches is evaluated.
Copy link

Thank you for submitting a Pull Request (PR) to the LLVM Project!

This PR will be automatically labeled and the relevant teams will be notified.

If you wish to, you can add reviewers by using the "Reviewers" section on this page.

If this is not working for you, it is probably because you do not have write permissions for the repository. In which case you can instead tag reviewers by name in a comment by using @ followed by their GitHub username.

If you have received no comments on your PR for a week, you can request a review by "ping"ing the PR by adding a comment “Ping”. The common courtesy "ping" rate is once a week. Please remember that you are asking for valuable time from other developers.

If you have further questions, they may be answered by the LLVM GitHub User Guide.

You can also ask questions in a comment on this PR, on the LLVM Discord or on the forums.

@JanJecmen
Copy link
Contributor Author

@aleks-tmb Hi, this is the separate PR for the LoopStructure change as requested in your review in #104659. Can I ask you to review this one as well, please?
Thanks!

@aleks-tmb
Copy link
Contributor

@JanJecmen LGTM!

@llvmbot
Copy link
Member

llvmbot commented Nov 16, 2024

@llvm/pr-subscribers-llvm-transforms

Author: Jan Ječmen (JanJecmen)

Changes

This refactoring hoists the profitability check earlier in the pipeline, so that for loops that are not profitable to transform there is no iteration over the basic blocks or LoopStructure computation.

Motivated by PR #104659 that tweaks how the profitability of individual branches is evaluated.


Full diff: https://github.com/llvm/llvm-project/pull/116384.diff

1 Files Affected:

  • (modified) llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp (+16-8)
diff --git a/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp b/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp
index a49dc7d30a00b0..0bc783412595e5 100644
--- a/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp
+++ b/llvm/lib/Transforms/Scalar/InductiveRangeCheckElimination.cpp
@@ -248,7 +248,7 @@ class InductiveRangeCheckElimination {
 
   // Returns true if it is profitable to do a transform basing on estimation of
   // number of iterations.
-  bool isProfitableToTransform(const Loop &L, LoopStructure &LS);
+  bool isProfitableToTransform(const Loop &L);
 
 public:
   InductiveRangeCheckElimination(ScalarEvolution &SE,
@@ -938,14 +938,12 @@ PreservedAnalyses IRCEPass::run(Function &F, FunctionAnalysisManager &AM) {
   return getLoopPassPreservedAnalyses();
 }
 
-bool
-InductiveRangeCheckElimination::isProfitableToTransform(const Loop &L,
-                                                        LoopStructure &LS) {
+bool InductiveRangeCheckElimination::isProfitableToTransform(const Loop &L) {
   if (SkipProfitabilityChecks)
     return true;
   if (GetBFI) {
     BlockFrequencyInfo &BFI = (*GetBFI)();
-    uint64_t hFreq = BFI.getBlockFreq(LS.Header).getFrequency();
+    uint64_t hFreq = BFI.getBlockFreq(L.getHeader()).getFrequency();
     uint64_t phFreq = BFI.getBlockFreq(L.getLoopPreheader()).getFrequency();
     if (phFreq != 0 && hFreq != 0 && (hFreq / phFreq < MinRuntimeIterations)) {
       LLVM_DEBUG(dbgs() << "irce: could not prove profitability: "
@@ -958,8 +956,17 @@ InductiveRangeCheckElimination::isProfitableToTransform(const Loop &L,
 
   if (!BPI)
     return true;
+
+  auto *Latch = L.getLoopLatch();
+  if (!Latch)
+    return true;
+  auto *LatchBr = dyn_cast<BranchInst>(Latch->getTerminator());
+  if (!LatchBr)
+    return true;
+  auto LatchBrExitIdx = LatchBr->getSuccessor(0) == L.getHeader() ? 1 : 0;
+
   BranchProbability ExitProbability =
-      BPI->getEdgeProbability(LS.Latch, LS.LatchBrExitIdx);
+      BPI->getEdgeProbability(Latch, LatchBrExitIdx);
   if (ExitProbability > BranchProbability(1, MinRuntimeIterations)) {
     LLVM_DEBUG(dbgs() << "irce: could not prove profitability: "
                       << "the exit probability is too big " << ExitProbability
@@ -982,6 +989,9 @@ bool InductiveRangeCheckElimination::run(
     return false;
   }
 
+  if (!isProfitableToTransform(*L))
+    return false;
+
   LLVMContext &Context = Preheader->getContext();
   SmallVector<InductiveRangeCheck, 16> RangeChecks;
   bool Changed = false;
@@ -1017,8 +1027,6 @@ bool InductiveRangeCheckElimination::run(
     return Changed;
   }
   LoopStructure LS = *MaybeLoopStructure;
-  if (!isProfitableToTransform(*L, LS))
-    return Changed;
   const SCEVAddRecExpr *IndVar =
       cast<SCEVAddRecExpr>(SE.getMinusSCEV(SE.getSCEV(LS.IndVarBase), SE.getSCEV(LS.IndVarStep)));
 

@aleks-tmb aleks-tmb merged commit 78db4e9 into llvm:main Dec 4, 2024
10 checks passed
Copy link

github-actions bot commented Dec 4, 2024

@JanJecmen Congratulations on having your first Pull Request (PR) merged into the LLVM Project!

Your changes will be combined with recent changes from other authors, then tested by our build bots. If there is a problem with a build, you may receive a report in an email or a comment on this PR.

Please check whether problems have been caused by your change specifically, as the builds can include changes from many authors. It is not uncommon for your change to be included in a build that fails due to someone else's changes, or infrastructure issues.

How to do this, and the rest of the post-merge process, is covered in detail here.

If your change does cause a problem, it may be reverted, or you can revert it yourself. This is a normal part of LLVM development. You can fix your changes and open a new PR to merge them again.

If you don't get any reports, no action is required from you. Your changes are working as expected, well done!

@llvm-ci
Copy link
Collaborator

llvm-ci commented Dec 4, 2024

LLVM Buildbot has detected a new failure on builder openmp-offload-libc-amdgpu-runtime running on omp-vega20-1 while building llvm at step 7 "Add check check-offload".

Full details are available at: https://lab.llvm.org/buildbot/#/builders/73/builds/9586

Here is the relevant piece of the build log for the reference
Step 7 (Add check check-offload) failure: 1200 seconds without output running [b'ninja', b'-j 32', b'check-offload'], attempting to kill
...
PASS: libomptarget :: x86_64-unknown-linux-gnu-LTO :: offloading/bug53727.cpp (947 of 960)
PASS: libomptarget :: x86_64-unknown-linux-gnu-LTO :: offloading/bug47654.cpp (948 of 960)
PASS: libomptarget :: x86_64-unknown-linux-gnu-LTO :: offloading/bug50022.cpp (949 of 960)
PASS: libomptarget :: x86_64-unknown-linux-gnu-LTO :: offloading/test_libc.cpp (950 of 960)
PASS: libomptarget :: x86_64-unknown-linux-gnu-LTO :: offloading/wtime.c (951 of 960)
PASS: libomptarget :: x86_64-unknown-linux-gnu :: offloading/std_complex_arithmetic.cpp (952 of 960)
PASS: libomptarget :: x86_64-unknown-linux-gnu :: offloading/bug49021.cpp (953 of 960)
PASS: libomptarget :: x86_64-unknown-linux-gnu-LTO :: offloading/complex_reduction.cpp (954 of 960)
PASS: libomptarget :: x86_64-unknown-linux-gnu-LTO :: offloading/bug49021.cpp (955 of 960)
PASS: libomptarget :: x86_64-unknown-linux-gnu-LTO :: offloading/std_complex_arithmetic.cpp (956 of 960)
command timed out: 1200 seconds without output running [b'ninja', b'-j 32', b'check-offload'], attempting to kill
process killed by signal 9
program finished with exit code -1
elapsedTime=1238.581690

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants