Testing builtin fast lzma2 in ROOT

cmake/modules/RootBenchOptions.cmake

@@ -4,4 +4,11 @@
 # # TBD: to introduce special function for options (similar to root.git)
 #----------------------------------------------------------------------------
 option(coverage OFF)
-option(rootbench-datafiles OFF)
+
+option(rootbench-datafiles OFF "Download files from root.cern.ch")
+
+option(experiment-datafiles OFF "Use for benchmarking CMS and ATLAS files (> 1.5 GB)")
+if(experiment-datafiles)
+    # We need to enable download of datafiles from oot.cern.ch
+    set(rootbench-datafiles ON CACHE BOOL "Download files from root.cern.ch" FORCE)
+endif()

root/io/io/CMakeLists.txt

@@ -7,10 +7,11 @@ if(ROOT_root7_FOUND) # using ROOT_root7_FOUND as a proxy for "c++ standard >= 14
   RB_ADD_GBENCHMARK(TFile_RDFSnapshot
     TFile_RDFSnapshot.cxx
     LABEL short
-  LIBRARIES Core RIO ROOTDataFrame Tree TreePlayer MathCore)
+    LIBRARIES Core RIO ROOTDataFrame Tree TreePlayer MathCore)
 endif()
 
-if(rootbench-datafiles)
+if(rootbench-datafiles AND experiment-datafiles)
+
   # FIXME:Compression tests needs to have libEvent.so which is a part of ROOT build
   # Preloading doesn't help...
   #if(TARGET Event)
@@ -21,22 +22,21 @@ if(rootbench-datafiles)
   #    DOWNLOAD_DATAFILES Event0-sample.root)
   #endif()
 
-  # FIXME: too long benchmarks, needs to be optimised
-  #RB_ADD_GBENCHMARK(CompressionBenchmarks_LHCb
-  #  TFile_LHCb_Benchmarks.cxx
-  #  LABEL short
-  #  LIBRARIES Core RIO
-  #  DOWNLOAD_DATAFILES lhcb_B2ppKK2011_md_noPIDstrip.root)
+  RB_ADD_GBENCHMARK(CompressionBenchmarks_LHCb
+    TFile_LHCb_Benchmarks.cxx
+    LABEL short
+    LIBRARIES Core RIO Tree
+    DOWNLOAD_DATAFILES lhcb_B2ppKK2011_md_noPIDstrip.root)
 
-  #RB_ADD_GBENCHMARK(CompressionBenchmarks_NanoAOD
-  #  TFile_NanoAOD_Benchmarks.cxx
-  #  LABEL short
-  #  LIBRARIES Core RIO
-  #  DOWNLOAD_DATAFILES Run2012B_DoubleMuParked.root)
+  RB_ADD_GBENCHMARK(CompressionBenchmarks_NanoAOD
+    TFile_NanoAOD_Benchmarks.cxx
+    LABEL short
+    LIBRARIES Core RIO Tree
+    DOWNLOAD_DATAFILES Run2012B_DoubleMuParked.root NanoAOD_DoubleMuon_CMS2011OpenData.root)
 
-  #RB_ADD_GBENCHMARK(CompressionBenchmarks_ATLAS
-  #  TFile_ATLAS_Benchmarks.cxx
-  #  LABEL short
-  #  LIBRARIES Core RIO
-  #  DOWNLOAD_DATAFILES gg_data-zstd.root)
+  RB_ADD_GBENCHMARK(CompressionBenchmarks_ATLAS
+    TFile_ATLAS_Benchmarks.cxx
+    LABEL short
+    LIBRARIES Core RIO Tree
+    DOWNLOAD_DATAFILES gg_data-zstd.root)
 endif()

root/io/io/TFile_ATLAS_Benchmarks.cxx

@@ -1,17 +1,20 @@
 #include "TFile.h"
 #include "TTree.h"
+#include "TStopwatch.h"
 
 #include "benchmark/benchmark.h"
 #include "rootbench/RBConfig.h"
 
 #include <map>
+#include <iostream>
 
 static std::string GetAlgoName(int algo) {
     std::map<int, std::string> algoName = {
         {1, "zlib"},
         {2, "lzma"},
         {4, "lz4"},
-        {5, "zstd"}
+        {5, "zstd"},
+        {6, "flzma2"}
     };
 
     if (algoName.find(algo) != algoName.end())
@@ -21,47 +24,57 @@ static std::string GetAlgoName(int algo) {
 }
 
 static void BM_ATLAS_Compress(benchmark::State &state, int algo) {
-    TFile *oldfile = new TFile((RB::GetDataDir() + "/gg_data-zstd.root").c_str());
-    TTree *oldtree = (TTree*)oldfile->Get("mini");
-
+    std::string path = RB::GetDataDir() + "/gg_data-zstd.root";
+    auto oldfile = TFile::Open(path.c_str());
+    auto oldtree = oldfile->Get<TTree>("mini");
+    TStopwatch timer;
+    const auto nevents = oldtree->GetEntries();
     int comp_level = state.range(0);
     std::string filename = "level_" + std::to_string(comp_level) + "_atlas_" + GetAlgoName(algo) + ".root";
 
     for (auto _ : state) {
         state.PauseTiming();
-
-        TFile *newfile = new TFile(filename.c_str(), "recreate");
-        TTree *newtree = oldtree->CloneTree();
+        auto newfile = new TFile(filename.c_str(), "recreate");
         newfile->SetCompressionAlgorithm(algo);
         newfile->SetCompressionLevel(comp_level);
+        auto newtree = oldtree->CloneTree();
 
         state.ResumeTiming();
+        timer.Start();
         newfile->Write();
+        timer.Stop();
         state.PauseTiming();
 
-        state.counters["comp_size"] = newfile->GetBytesWritten();
+        state.counters["comp_size"] = newfile->GetSize();
+        double rtime = timer.RealTime();
+        double ctime = timer.CpuTime();
+        state.counters["mb_rts"] = newfile->GetSize()/rtime;
+        state.counters["mb_cts"] = newfile->GetSize()/ctime;
         newfile->Close();
-
         state.ResumeTiming();
     }
 }
 
 static void BM_ATLAS_Decompress(benchmark::State &state, int algo) {
     int comp_level = state.range(0);
-
+    int nb;
+    TStopwatch timer;
     std::string filename = "level_" + std::to_string(comp_level) + "_atlas_" + GetAlgoName(algo) + ".root";
     for (auto _ : state) {
-        TFile *hfile = new TFile(filename.c_str());
-        TTree *tree = (TTree*)hfile->Get("mini");
-
-        Int_t nevent = (Int_t)tree->GetEntries();
-
-        Int_t nb = 0;
+        timer.Start();
+        TFile f(filename.c_str());
+        auto tree = static_cast<TTree *>(f.Get("mini"));
+        const auto nevents = tree->GetEntries();
         Int_t ev;
-
-        for (ev = 0; ev < nevent; ev++) {
+        for (ev = 0; ev < nevents; ++ev) {
             nb += tree->GetEntry(ev);
         }
+        timer.Stop();
+        double rtime = timer.RealTime();
+        double ctime = timer.CpuTime();
+        state.counters["mb_rts"] = f.GetSize()/rtime;
+        state.counters["mb_cts"] = f.GetSize()/ctime;
+        f.Close();
     }
 }
 
@@ -77,6 +90,9 @@ static void BM_ATLAS_Compress_LZ4(benchmark::State &state) {
 static void BM_ATLAS_Compress_ZSTD(benchmark::State &state) {
     BM_ATLAS_Compress(state, 5);
 }
+static void BM_ATLAS_Compress_FLZMA2(benchmark::State &state) {
+    BM_ATLAS_Compress(state, 6);
+}
 
 static void BM_ATLAS_Decompress_ZLIB(benchmark::State &state) {
     BM_ATLAS_Decompress(state, 1);
@@ -90,40 +106,51 @@ static void BM_ATLAS_Decompress_LZ4(benchmark::State &state) {
 static void BM_ATLAS_Decompress_ZSTD(benchmark::State &state) {
     BM_ATLAS_Decompress(state, 5);
 }
-
+static void BM_ATLAS_Decompress_FLZMA2(benchmark::State &state) {
+    BM_ATLAS_Decompress(state, 6);
+}
 
 BENCHMARK(BM_ATLAS_Compress_ZLIB)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(1);
 
 BENCHMARK(BM_ATLAS_Compress_LZMA)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(1);
 
 BENCHMARK(BM_ATLAS_Compress_LZ4)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(1);
 
 BENCHMARK(BM_ATLAS_Compress_ZSTD)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(1);
+
+BENCHMARK(BM_ATLAS_Compress_FLZMA2)
+->Arg(1)->Arg(6)->Arg(9)
+->Unit(benchmark::kMillisecond)->Iterations(1);
+
 
 
 BENCHMARK(BM_ATLAS_Decompress_ZLIB)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(1);
 
 BENCHMARK(BM_ATLAS_Decompress_LZMA)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(1);
 
 BENCHMARK(BM_ATLAS_Decompress_LZ4)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(1);
 
 BENCHMARK(BM_ATLAS_Decompress_ZSTD)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(1);
+
+BENCHMARK(BM_ATLAS_Decompress_FLZMA2)
+->Arg(1)->Arg(6)->Arg(9)
+->Unit(benchmark::kMillisecond)->Iterations(1);
 
 
 BENCHMARK_MAIN(); 

root/io/io/TFile_CompressionBenchmarks_MainEvent.cxx

@@ -13,7 +13,8 @@ static std::string GetAlgoName(int algo) {
       {1, "zlib"},
       {2, "lzma"},
       {4, "lz4"},
-      {5, "zstd"}
+      {5, "zstd"},
+      {6, "flzma2"}
    };
 
    if (algoName.find(algo) != algoName.end())
@@ -87,6 +88,9 @@ static void BM_MainEvent_Compress_LZ4(benchmark::State &state) {
 static void BM_MainEvent_Compress_ZSTD(benchmark::State &state) {
    BM_MainEvent_Compress(state, 5);
 }
+static void BM_MainEvent_Compress_FLZMA2(benchmark::State &state) {
+   BM_MainEvent_Compress(state, 6);
+}
 
 
 static void BM_MainEvent_Decompress_ZLIB(benchmark::State &state) {
@@ -101,7 +105,9 @@ static void BM_MainEvent_Decompress_LZ4(benchmark::State &state) {
 static void BM_MainEvent_Decompress_ZSTD(benchmark::State &state) {
    BM_MainEvent_Decompress(state, 5);
 }
-
+static void BM_MainEvent_Decompress_FLZMA2(benchmark::State &state) {
+   BM_MainEvent_Decompress(state, 6);
+}
 
 
 BENCHMARK(BM_MainEvent_Compress_ZLIB)
@@ -120,6 +126,10 @@ BENCHMARK(BM_MainEvent_Compress_ZSTD)
 ->Arg(1)->Arg(6)->Arg(9)
 ->Unit(benchmark::kMillisecond)->Iterations(5);
 
+BENCHMARK(BM_MainEvent_Compress_FLZMA2)
+->Arg(1)->Arg(6)->Arg(9)
+->Unit(benchmark::kMillisecond)->Iterations(5);
+
 
 BENCHMARK(BM_MainEvent_Decompress_ZLIB)
 ->Arg(1)->Arg(6)->Arg(9)
@@ -137,5 +147,10 @@ BENCHMARK(BM_MainEvent_Decompress_ZSTD)
 ->Arg(1)->Arg(6)->Arg(9)
 ->Unit(benchmark::kMillisecond)->Iterations(5);
 
+BENCHMARK(BM_MainEvent_Decompress_FLZMA2)
+->Arg(1)->Arg(6)->Arg(9)
+->Unit(benchmark::kMillisecond)->Iterations(5);
+
+
 
 BENCHMARK_MAIN();