[llvm] [MCSched] Allow tuning LoadLatency/MispredictPenalty (PR #203139)

[via llvm-commits]

github-actions[bot] wrote:

<!--LLVM CODE FORMAT COMMENT: {clang-format}-->


:warning: C/C++ code formatter, clang-format found issues in your code. :warning:

<details>
<summary>
You can test this locally with the following command:
</summary>

``````````bash
git-clang-format --diff origin/main HEAD --extensions h,cpp -- llvm/include/llvm/MC/MCSchedule.h llvm/lib/CodeGen/EarlyIfConversion.cpp llvm/lib/CodeGen/SelectOptimize.cpp llvm/lib/CodeGen/TargetInstrInfo.cpp llvm/lib/MC/MCSchedule.cpp llvm/lib/Target/AArch64/AArch64ConditionalCompares.cpp llvm/lib/Target/ARM/ARMSubtarget.cpp llvm/lib/Target/RISCV/RISCVSubtarget.cpp llvm/lib/Target/X86/X86CmovConversion.cpp --diff_from_common_commit
``````````

:warning:
The reproduction instructions above might return results for more than one PR
in a stack if you are using a stacked PR workflow. You can limit the results by
changing `origin/main` to the base branch/commit you want to compare against.
:warning:

</details>

<details>
<summary>
View the diff from clang-format here.
</summary>

``````````diff
diff --git a/llvm/lib/CodeGen/SelectOptimize.cpp b/llvm/lib/CodeGen/SelectOptimize.cpp
index 2637e6478..01d69faaa 100644
--- a/llvm/lib/CodeGen/SelectOptimize.cpp
+++ b/llvm/lib/CodeGen/SelectOptimize.cpp
@@ -1414,7 +1414,8 @@ SelectOptimizeImpl::computeInstCost(const Instruction *I) {
 ScaledNumber<uint64_t>
 SelectOptimizeImpl::getMispredictionCost(const SelectLike SI,
                                          const Scaled64 CondCost) {
-  uint64_t MispredictPenalty = TSchedModel.getMCSchedModel()->getMispredictPenalty();
+  uint64_t MispredictPenalty =
+      TSchedModel.getMCSchedModel()->getMispredictPenalty();
 
   // Account for the default misprediction rate when using a branch
   // (conservatively set to 25% by default).
diff --git a/llvm/lib/MC/MCSchedule.cpp b/llvm/lib/MC/MCSchedule.cpp
index 0aa7f8e6f..8808343b2 100644
--- a/llvm/lib/MC/MCSchedule.cpp
+++ b/llvm/lib/MC/MCSchedule.cpp
@@ -34,8 +34,7 @@ static cl::opt<int> SchedMispredictPenalty(
              "model. A non-negative value overrides the target default."));
 
 static cl::opt<int> SchedLoadLatency(
-    "sched-load-latency", cl::Hidden, cl::init(-1),
-    cl::cat(MCScheduleOptions),
+    "sched-load-latency", cl::Hidden, cl::init(-1), cl::cat(MCScheduleOptions),
     cl::desc("Override the load latency (in cycles) in the scheduler model. "
              "A non-negative value overrides the target default."));
 
diff --git a/llvm/lib/Target/X86/X86CmovConversion.cpp b/llvm/lib/Target/X86/X86CmovConversion.cpp
index 25d23b547..a14d8d897 100644
--- a/llvm/lib/Target/X86/X86CmovConversion.cpp
+++ b/llvm/lib/Target/X86/X86CmovConversion.cpp
@@ -546,7 +546,8 @@ bool X86CmovConversionImpl::checkForProfitableCmovCandidates(
   //   To be conservative, the gain of such CMOV transformation should cover at
   //   at least 25% of branch-misprediction-penalty.
   //===--------------------------------------------------------------------===//
-  unsigned MispredictPenalty = TSchedModel.getMCSchedModel()->getMispredictPenalty();
+  unsigned MispredictPenalty =
+      TSchedModel.getMCSchedModel()->getMispredictPenalty();
   CmovGroups TempGroups;
   std::swap(TempGroups, CmovInstGroups);
   for (auto &Group : TempGroups) {

``````````

</details>


https://github.com/llvm/llvm-project/pull/203139