Better configuration of RNTupleTemp*

Dr15Jones · Dr15Jones · commit 2553a20deed5 · 2025-12-02T15:04:47.000-06:00
- made TTree only parameters optional
- added RNTuple specific options
diff --git a/FWIO/RNTupleTempInput/src/EmbeddedRNTupleTempSource.cc b/FWIO/RNTupleTempInput/src/EmbeddedRNTupleTempSource.cc
@@ -14,6 +14,14 @@ namespace edm {
   class EventPrincipal;
 }  // namespace edm
 namespace edm::rntuple_temp {
+  namespace {
+    EmbeddedRNTupleTempSource::Optimizations fromConfig(edm::ParameterSet const& iConfig) {
+      EmbeddedRNTupleTempSource::Optimizations opts;
+      opts.useClusterCache = iConfig.getUntrackedParameter<bool>("useClusterCache");
+      return opts;
+    }
+
+  }  // namespace
 
   EmbeddedRNTupleTempSource::EmbeddedRNTupleTempSource(ParameterSet const& pset,
                                                        VectorInputSourceDescription const& desc)
@@ -30,7 +38,8 @@ namespace edm::rntuple_temp {
         //
         skipBadFiles_(pset.getUntrackedParameter<bool>("skipBadFiles", false)),
         bypassVersionCheck_(pset.getUntrackedParameter<bool>("bypassVersionCheck", false)),
-        treeMaxVirtualSize_(pset.getUntrackedParameter<int>("treeMaxVirtualSize", -1)),
+        treeMaxVirtualSize_(0),
+        optimizations_(fromConfig(pset.getUntrackedParameterSet("rntupleReadOptions"))),
         productSelectorRules_(pset, "inputCommands", "InputSource"),
         runHelper_(new DefaultInputSourceRunHelper()),
         catalog_(pset.getUntrackedParameter<std::vector<std::string> >("fileNames"),
@@ -91,8 +100,13 @@ namespace edm::rntuple_temp {
         ->setComment(
             "True:  Bypass release version check.\n"
             "False: Throw exception if reading file in a release prior to the release in which the file was written.");
-    desc.addUntracked<int>("treeMaxVirtualSize", -1)
-        ->setComment("Size of ROOT TTree TBasket cache.  Affects performance.");
+    desc.addUntracked<int>("treeMaxVirtualSize", -1)->setComment("Not used by RNTuple.");
+    {
+      ParameterSetDescription rntupleReadOptions;
+      rntupleReadOptions.addUntracked<bool>("useClusterCache", true)
+          ->setComment("True: use ROOT cluster cache. False: do not use cluster cache.");
+      desc.addUntracked("rntupleReadOptions", rntupleReadOptions);
+    }
 
     ProductSelectorRules::fillDescription(desc, "inputCommands");
     RootEmbeddedFileSequence::fillDescription(desc);
diff --git a/FWIO/RNTupleTempInput/src/EmbeddedRNTupleTempSource.h b/FWIO/RNTupleTempInput/src/EmbeddedRNTupleTempSource.h
@@ -35,6 +35,10 @@ namespace edm::rntuple_temp {
 
   class EmbeddedRNTupleTempSource : public VectorInputSource {
   public:
+    struct Optimizations {
+      bool useClusterCache = true;
+    };
+
     explicit EmbeddedRNTupleTempSource(ParameterSet const& pset, VectorInputSourceDescription const& desc);
     ~EmbeddedRNTupleTempSource() override;
     using VectorInputSource::processHistoryRegistryForUpdate;
@@ -45,6 +49,7 @@ namespace edm::rntuple_temp {
     bool bypassVersionCheck() const { return bypassVersionCheck_; }
     unsigned int nStreams() const { return nStreams_; }
     int treeMaxVirtualSize() const { return treeMaxVirtualSize_; }
+    Optimizations const& optimizations() const { return optimizations_; }
     ProductSelectorRules const& productSelectorRules() const { return productSelectorRules_; }
     InputSourceRunHelperBase* runHelper() { return runHelper_.get(); }
 
@@ -68,6 +73,7 @@ namespace edm::rntuple_temp {
     bool skipBadFiles_;
     bool bypassVersionCheck_;
     int const treeMaxVirtualSize_;
+    Optimizations optimizations_;
     ProductSelectorRules productSelectorRules_;
     std::unique_ptr<InputSourceRunHelperBase> runHelper_;
 
diff --git a/FWIO/RNTupleTempInput/src/RNTupleTempSource.cc b/FWIO/RNTupleTempInput/src/RNTupleTempSource.cc
@@ -62,6 +62,11 @@ namespace edm::rntuple_temp {
             << primary.id() << " has inconsistent RunAuxiliary data in the primary and secondary file\n";
       }
     }
+    RNTupleTempSource::Optimizations fromConfig(edm::ParameterSet const& iConfig) {
+      RNTupleTempSource::Optimizations opts;
+      opts.useClusterCache = iConfig.getUntrackedParameter<bool>("useClusterCache");
+      return opts;
+    }
   }  // namespace
 
   RNTupleTempSource::RNTupleTempSource(ParameterSet const& pset, InputSourceDescription const& desc)
@@ -79,7 +84,8 @@ namespace edm::rntuple_temp {
         nStreams_(desc.allocations_->numberOfStreams()),
         skipBadFiles_(pset.getUntrackedParameter<bool>("skipBadFiles")),
         bypassVersionCheck_(pset.getUntrackedParameter<bool>("bypassVersionCheck")),
-        treeMaxVirtualSize_(pset.getUntrackedParameter<int>("treeMaxVirtualSize")),
+        treeMaxVirtualSize_(0),
+        optimizations_(fromConfig(pset.getUntrackedParameterSet("rntupleReadOptions"))),
         productSelectorRules_(pset, "inputCommands", "InputSource"),
         dropDescendants_(pset.getUntrackedParameter<bool>("dropDescendantsOfDroppedBranches")),
         labelRawDataLikeMC_(pset.getUntrackedParameter<bool>("labelRawDataLikeMC")),
@@ -301,7 +307,7 @@ namespace edm::rntuple_temp {
     ParameterSetDescription desc;
 
     std::vector<std::string> defaultStrings;
-    desc.setComment("Reads EDM/Root files.");
+    desc.setComment("Reads EDM/RNTuple Root files.");
     desc.addUntracked<std::vector<std::string> >("fileNames")->setComment("Names of files to be processed.");
     desc.addUntracked<std::vector<std::string> >("secondaryFileNames", defaultStrings)
         ->setComment("Names of secondary files to be processed.");
@@ -316,8 +322,13 @@ namespace edm::rntuple_temp {
         ->setComment(
             "True:  Bypass release version check.\n"
             "False: Throw exception if reading file in a release prior to the release in which the file was written.");
-    desc.addUntracked<int>("treeMaxVirtualSize", -1)
-        ->setComment("Size of ROOT TTree TBasket cache. Affects performance.");
+    desc.addOptionalUntracked<int>("treeMaxVirtualSize", -1)->setComment("Not used by RNTuple.");
+    {
+      ParameterSetDescription rntupleReadOptions;
+      rntupleReadOptions.addUntracked<bool>("useClusterCache", true)
+          ->setComment("True: use ROOT cluster cache. False: do not use cluster cache.");
+      desc.addUntracked("rntupleReadOptions", rntupleReadOptions);
+    }
     desc.addUntracked<bool>("dropDescendantsOfDroppedBranches", true)
         ->setComment("If True, also drop on input any descendent of any branch dropped on input.");
     desc.addUntracked<bool>("labelRawDataLikeMC", true)
diff --git a/FWIO/RNTupleTempInput/src/RNTupleTempSource.h b/FWIO/RNTupleTempInput/src/RNTupleTempSource.h
@@ -32,6 +32,9 @@ namespace edm::rntuple_temp {
 
   class RNTupleTempSource : public InputSource {
   public:
+    struct Optimizations {
+      bool useClusterCache = true;
+    };
     explicit RNTupleTempSource(ParameterSet const& pset, InputSourceDescription const& desc);
     ~RNTupleTempSource() override;
     using InputSource::processHistoryRegistryForUpdate;
@@ -45,6 +48,7 @@ namespace edm::rntuple_temp {
     bool delayReadingEventProducts() const { return delayReadingEventProducts_; }
     unsigned int nStreams() const { return nStreams_; }
     int treeMaxVirtualSize() const { return treeMaxVirtualSize_; }
+    Optimizations const& optimizations() const { return optimizations_; }
     ProductSelectorRules const& productSelectorRules() const { return productSelectorRules_; }
     InputSourceRunHelperBase* runHelper() { return runHelper_.get(); }
 
@@ -87,6 +91,7 @@ namespace edm::rntuple_temp {
     bool skipBadFiles_;
     bool bypassVersionCheck_;
     int const treeMaxVirtualSize_;
+    Optimizations optimizations_;
     ProductSelectorRules productSelectorRules_;
     bool dropDescendants_;
     bool labelRawDataLikeMC_;
diff --git a/FWIO/RNTupleTempInput/src/RootEmbeddedFileSequence.cc b/FWIO/RNTupleTempInput/src/RootEmbeddedFileSequence.cc
@@ -144,9 +144,7 @@ namespace edm::rntuple_temp {
                                                             .enforceGUIDInFileName = enforceGUIDInFileName_},
                                       InputType::SecondarySource,
                                       RootFile::ProcessingOptions{},
-                                      RootFile::TTreeOptions{.treeCacheSize = treeCacheSize_,
-                                                             .treeMaxVirtualSize = input_.treeMaxVirtualSize(),
-                                                             .enablePrefetching = enablePrefetching_},
+                                      RootFile::TTreeOptions{.useClusterCache = input_.optimizations().useClusterCache},
                                       RootFile::ProductChoices{.productSelectorRules = input_.productSelectorRules()},
                                       RootFile::CrossFileInfo{.runHelper = input_.runHelper(),
                                                               .indexesIntoFiles = indexesIntoFiles(),
@@ -379,8 +377,7 @@ namespace edm::rntuple_temp {
         ->setComment(
             "How many files to try if 'sequential' is False and 'skipBadFiles' is True.\n"
             "Defaults to 3 (or # of files if smaller).");
-    desc.addUntracked<unsigned int>("cacheSize", rootrntuple::defaultCacheSize)
-        ->setComment("Size of ROOT TTree prefetch cache.  Affects performance.");
+    desc.addOptionalUntracked<unsigned int>("cacheSize", 0)->setComment("Not used by RNTuple.");
     desc.addUntracked<bool>("enforceGUIDInFileName", false)
         ->setComment(
             "True:  file name part is required to be equal to the GUID of the file\n"
diff --git a/FWIO/RNTupleTempInput/src/RootFile.cc b/FWIO/RNTupleTempInput/src/RootFile.cc
@@ -169,20 +169,9 @@ namespace edm::rntuple_temp {
         hasNewlyDroppedBranch_(),
         branchListIndexesUnchanged_(false),
         eventAuxCache_(),
-        eventTree_(
-            fileOptions.filePtr, InEvent, nStreams, ttreeOptions, rootrntuple::defaultLearningEntries, inputType),
-        lumiTree_(fileOptions.filePtr,
-                  InLumi,
-                  1,
-                  ttreeOptions.usingDefaultNonEventOptions(),
-                  rootrntuple::defaultNonEventLearningEntries,
-                  inputType),
-        runTree_(fileOptions.filePtr,
-                 InRun,
-                 1,
-                 ttreeOptions.usingDefaultNonEventOptions(),
-                 rootrntuple::defaultNonEventLearningEntries,
-                 inputType),
+        eventTree_(fileOptions.filePtr, InEvent, nStreams, ttreeOptions, inputType),
+        lumiTree_(fileOptions.filePtr, InLumi, 1, ttreeOptions.usingDefaultNonEventOptions(), inputType),
+        runTree_(fileOptions.filePtr, InRun, 1, ttreeOptions.usingDefaultNonEventOptions(), inputType),
         treePointers_(),
         lastEventEntryNumberRead_(IndexIntoFile::invalidEntry),
         productRegistry_(),
@@ -1895,13 +1884,8 @@ namespace edm::rntuple_temp {
     // the ProcessBlock TTree's in the file. (later in the RootFile constructor, dropOnInput might
     // remove some and also reordering may occur).
     for (auto const& process : storedProcessBlockHelper.processesWithProcessBlockProducts()) {
-      processBlockTrees_.emplace_back(std::make_unique<RootRNTuple>(filePtr,
-                                                                    InProcess,
-                                                                    process,
-                                                                    1,
-                                                                    options.usingDefaultNonEventOptions(),
-                                                                    rootrntuple::defaultNonEventLearningEntries,
-                                                                    inputType));
+      processBlockTrees_.emplace_back(std::make_unique<RootRNTuple>(
+          filePtr, InProcess, process, 1, options.usingDefaultNonEventOptions(), inputType));
     }
   }
 
diff --git a/FWIO/RNTupleTempInput/src/RootPrimaryFileSequence.cc b/FWIO/RNTupleTempInput/src/RootPrimaryFileSequence.cc
@@ -32,7 +32,7 @@ namespace edm::rntuple_temp {
         initialNumberOfEventsToSkip_(pset.getUntrackedParameter<unsigned int>("skipEvents")),
         noRunLumiSort_(pset.getUntrackedParameter<bool>("noRunLumiSort")),
         noEventSort_(noRunLumiSort_ ? true : pset.getUntrackedParameter<bool>("noEventSort")),
-        treeCacheSize_(noEventSort_ ? pset.getUntrackedParameter<unsigned int>("cacheSize") : 0U),
+        treeCacheSize_(0U),
         duplicateChecker_(new DuplicateChecker(pset)),
         usingGoToEvent_(false),
         enablePrefetching_(false),
@@ -161,9 +161,7 @@ namespace edm::rntuple_temp {
                                     .noRunLumiSort = noRunLumiSort_,
                                     .noEventSort = noEventSort_,
                                     .usingGoToEvent = usingGoToEvent_},
-        RootFile::TTreeOptions{.treeCacheSize = treeCacheSize_,
-                               .treeMaxVirtualSize = input_.treeMaxVirtualSize(),
-                               .enablePrefetching = enablePrefetching_,
+        RootFile::TTreeOptions{.useClusterCache = input_.optimizations().useClusterCache,
                                .promptReading = not input_.delayReadingEventProducts()},
         RootFile::ProductChoices{.productSelectorRules = input_.productSelectorRules(),
                                  .associationsFromSecondary = nullptr,  // associationsFromSecondary
@@ -427,8 +425,7 @@ namespace edm::rntuple_temp {
         ->setComment(
             "True:  Process runs, lumis and events in the order they appear in the file.\n"
             "False: Follow settings based on 'noEventSort' setting.");
-    desc.addUntracked<unsigned int>("cacheSize", rootrntuple::defaultCacheSize)
-        ->setComment("Size of ROOT TTree prefetch cache.  Affects performance.");
+    desc.addOptionalUntracked<unsigned int>("cacheSize", 0)->setComment("Not used by RNTuple");
     std::string defaultString("permissive");
     desc.addUntracked<std::string>("branchesMustMatch", defaultString)
         ->setComment(
diff --git a/FWIO/RNTupleTempInput/src/RootRNTuple.cc b/FWIO/RNTupleTempInput/src/RootRNTuple.cc
@@ -35,14 +35,11 @@ namespace edm::rntuple_temp {
   RootRNTuple::RootRNTuple(std::shared_ptr<InputFile> filePtr,
                            BranchType const& branchType,
                            unsigned int nIndexes,
-                           unsigned int learningEntries,
-                           bool enablePrefetching,
                            bool promptRead,
                            InputType inputType)
       : filePtr_(filePtr),
         branchType_(branchType),
         entryNumberForIndex_(std::make_unique<std::vector<EntryNumber>>(nIndexes, IndexIntoFile::invalidEntry)),
-        enablePrefetching_(enablePrefetching),
         promptRead_(promptRead),
         rootDelayedReader_(makeRootDelayedReader(*this, filePtr, inputType, nIndexes, promptRead)) {}
 
@@ -51,11 +48,9 @@ namespace edm::rntuple_temp {
                            BranchType const& branchType,
                            unsigned int nIndexes,
                            Options const& options,
-                           unsigned int learningEntries,
                            InputType inputType)
-      : RootRNTuple(
-            filePtr, branchType, nIndexes, learningEntries, options.enablePrefetching, options.promptReading, inputType) {
-    init(BranchTypeToProductTreeName(branchType), options.treeMaxVirtualSize, options.treeCacheSize);
+      : RootRNTuple(filePtr, branchType, nIndexes, options.promptReading, inputType) {
+    init(BranchTypeToProductTreeName(branchType), options);
     auxDesc_ = getAuxiliaryFieldId(*reader_, branchType_);
   }
 
@@ -65,19 +60,20 @@ namespace edm::rntuple_temp {
                            std::string const& processName,
                            unsigned int nIndexes,
                            Options const& options,
-                           unsigned int learningEntries,
                            InputType inputType)
-      : RootRNTuple(
-            filePtr, branchType, nIndexes, learningEntries, options.enablePrefetching, options.promptReading, inputType) {
+      : RootRNTuple(filePtr, branchType, nIndexes, options.promptReading, inputType) {
     processName_ = processName;
-    init(BranchTypeToProductTreeName(branchType, processName), options.treeMaxVirtualSize, options.treeCacheSize);
+    init(BranchTypeToProductTreeName(branchType, processName), options);
   }
 
-  void RootRNTuple::init(std::string const& productTreeName, unsigned int maxVirtualSize, unsigned int cacheSize) {
+  void RootRNTuple::init(std::string const& productTreeName, Options const& options) {
     if (filePtr_.get() != nullptr) {
       auto tuple = filePtr_->Get<ROOT::RNTuple>(productTreeName.c_str());
       if (tuple != nullptr) {
-        reader_ = ROOT::RNTupleReader::Open(*tuple);
+        ROOT::RNTupleReadOptions rntupleOptions;
+        rntupleOptions.SetClusterCache(options.useClusterCache ? ROOT::RNTupleReadOptions::EClusterCache::kOn
+                                                               : ROOT::RNTupleReadOptions::EClusterCache::kOff);
+        reader_ = ROOT::RNTupleReader::Open(*tuple, rntupleOptions);
       }
     }
     if (not reader_) {
@@ -86,9 +82,6 @@ namespace edm::rntuple_temp {
           << "\n This is either not an edm ROOT file or is one that has been corrupted.";
     }
     entries_ = reader_->GetNEntries();
-
-    setTreeMaxVirtualSize(maxVirtualSize);
-    setCacheSize(cacheSize);
   }
 
   RootRNTuple::~RootRNTuple() {}
diff --git a/FWIO/RNTupleTempInput/src/RootRNTuple.h b/FWIO/RNTupleTempInput/src/RootRNTuple.h
@@ -36,10 +36,6 @@ namespace edm::rntuple_temp {
   class RootDelayedReaderBase;
 
   namespace rootrntuple {
-    unsigned int const defaultCacheSize = 20U * 1024 * 1024;
-    unsigned int const defaultNonEventCacheSize = 1U * 1024 * 1024;
-    unsigned int const defaultLearningEntries = 20U;
-    unsigned int const defaultNonEventLearningEntries = 1U;
     using EntryNumber = IndexIntoFile::EntryNumber_t;
     struct ProductInfo {
       ProductInfo(ProductDescription const& prod) : productDescription_(prod) {}
@@ -94,33 +90,25 @@ namespace edm::rntuple_temp {
     using ProductMap = rootrntuple::ProductMap;
     using EntryNumber = rootrntuple::EntryNumber;
     struct Options {
-      unsigned int treeCacheSize = 0U;
-      int treeMaxVirtualSize;
-      bool enablePrefetching;
+      bool useClusterCache = true;
       bool promptReading = false;
 
-      Options usingDefaultNonEventOptions() const {
-        return {rootrntuple::defaultNonEventCacheSize, treeMaxVirtualSize, enablePrefetching, false};
-      }
+      Options usingDefaultNonEventOptions() const { return {}; }
     };
 
     RootRNTuple(std::shared_ptr<InputFile> filePtr,
                 BranchType const& branchType,
                 unsigned int nIndexes,
                 Options const& options,
-                unsigned int learningEntries,
                 InputType inputType);
 
     RootRNTuple(std::shared_ptr<InputFile> filePtr,
                 BranchType const& branchType,
                 std::string const& processName,
                 unsigned int nIndexes,
                 Options const& options,
-                unsigned int learningEntries,
                 InputType inputType);
 
-    void init(std::string const& productTreeName, unsigned int maxVirtualSize, unsigned int cacheSize);
-
     ~RootRNTuple();
 
     RootRNTuple(RootRNTuple const&) = delete;             // Disallow copying and moving
@@ -185,11 +173,11 @@ namespace edm::rntuple_temp {
         signalslot::Signal<void(StreamContext const&, ModuleCallingContext const&)> const* postEventReadSource);
 
   private:
+    void init(std::string const& productTreeName, Options const& options);
+
     RootRNTuple(std::shared_ptr<InputFile> filePtr,
                 BranchType const& branchType,
                 unsigned int nIndexes,
-                unsigned int learningEntries,
-                bool enablePrefetching,
                 bool promptRead,
                 InputType inputType);
 
@@ -211,7 +199,6 @@ namespace edm::rntuple_temp {
     ProductMap branches_;
     unsigned int cacheSize_ = 0;
     unsigned long treeAutoFlush_ = 0;
-    bool enablePrefetching_;
     bool promptRead_;
     std::unique_ptr<RootDelayedReaderBase> rootDelayedReader_;
   };
diff --git a/FWIO/RNTupleTempInput/src/RootSecondaryFileSequence.cc b/FWIO/RNTupleTempInput/src/RootSecondaryFileSequence.cc
@@ -82,8 +82,7 @@ namespace edm::rntuple_temp {
         RootFile::ProcessingOptions{
             .processingMode = input_.processingMode(),
         },
-        RootFile::TTreeOptions{.treeMaxVirtualSize = input_.treeMaxVirtualSize(),
-                               .enablePrefetching = enablePrefetching_,
+        RootFile::TTreeOptions{.useClusterCache = input_.optimizations().useClusterCache,
                                .promptReading = not input_.delayReadingEventProducts()},
         RootFile::ProductChoices{.productSelectorRules = input_.productSelectorRules(),
                                  .associationsFromSecondary = &associationsFromSecondary_,
diff --git a/FWIO/RNTupleTempOutput/interface/RNTupleTempOutputModule.h b/FWIO/RNTupleTempOutput/interface/RNTupleTempOutputModule.h
diff --git a/FWIO/RNTupleTempOutput/src/RNTupleTempOutputModule.cc b/FWIO/RNTupleTempOutput/src/RNTupleTempOutputModule.cc
diff --git a/FWIO/RNTupleTempOutput/src/RootOutputFile.cc b/FWIO/RNTupleTempOutput/src/RootOutputFile.cc
diff --git a/FWIO/RNTupleTempOutput/src/RootOutputRNTuple.cc b/FWIO/RNTupleTempOutput/src/RootOutputRNTuple.cc
diff --git a/FWIO/RNTupleTempOutput/src/RootOutputRNTuple.h b/FWIO/RNTupleTempOutput/src/RootOutputRNTuple.h
