Fix compression benchmarks

root/io/io/CMakeLists.txt

@@ -15,22 +15,21 @@ if(rootbench-datafiles)
     LIBRARIES Core RIO
     DOWNLOAD_DATAFILES Event0-sample.root)
 
-  # 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
+    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
+    DOWNLOAD_DATAFILES Run2012B_DoubleElectron.root)
 
-  #RB_ADD_GBENCHMARK(CompressionBenchmarks_ATLAS
-  #  TFile_ATLAS_Benchmarks.cxx
-  #  LABEL short
-  #  LIBRARIES Core RIO
-  #  DOWNLOAD_DATAFILES gg_data-zstd.root)
-endif()
+  RB_ADD_GBENCHMARK(CompressionBenchmarks_ATLAS
+    TFile_ATLAS_Benchmarks.cxx
+    LABEL short
+    LIBRARIES Core RIO
+    DOWNLOAD_DATAFILES atlasopendata_mc_117050.ttbar_lep.root)
+endif()

root/io/io/TFile_ATLAS_Benchmarks.cxx

@@ -1,5 +1,6 @@
 #include "TFile.h"
 #include "TTree.h"
+#include "TSystem.h"
 
 #include "benchmark/benchmark.h"
 #include "rootbench/RBConfig.h"
@@ -20,38 +21,22 @@ static std::string GetAlgoName(int algo) {
         return "error";
 }
 
-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");
+static void BM_ATLAS_Decompress(benchmark::State &state, int algo, std::string filename) {
 
     int comp_level = state.range(0);
-    std::string filename = "level_" + std::to_string(comp_level) + "_atlas_" + GetAlgoName(algo) + ".root";
+    std::string comp_setting = std::to_string(algo * 100 + comp_level);
+    std::string old_filename = (RB::GetDataDir() + "/" + filename + ".root").c_str();
+    std::string new_filename = (RB::GetTempFs() + "level_" + std::to_string(comp_level) + "_atlas_" + GetAlgoName(algo) + ".root").c_str();
 
-    for (auto _ : state) {
-        state.PauseTiming();
-
-        TFile *newfile = new TFile(filename.c_str(), "recreate");
-        TTree *newtree = oldtree->CloneTree();
-        newfile->SetCompressionAlgorithm(algo);
-        newfile->SetCompressionLevel(comp_level);
+    gSystem->Exec(("hadd -v 0 -f" + comp_setting + " " + new_filename + " " + old_filename).c_str());
 
-        state.ResumeTiming();
-        newfile->Write();
-        state.PauseTiming();
+    TFile *newfile = new TFile(new_filename.c_str());
+    state.counters["comp_size"] = newfile->GetSize();
+    newfile->Close();
 
-        state.counters["comp_size"] = newfile->GetBytesWritten();
-        newfile->Close();
-
-        state.ResumeTiming();
-    }
-}
-
-static void BM_ATLAS_Decompress(benchmark::State &state, int algo) {
-    int comp_level = state.range(0);
-
-    std::string filename = "level_" + std::to_string(comp_level) + "_atlas_" + GetAlgoName(algo) + ".root";
     for (auto _ : state) {
-        TFile *hfile = new TFile(filename.c_str());
+
+        TFile *hfile = new TFile(new_filename.c_str());
         TTree *tree = (TTree*)hfile->Get("mini");
 
         Int_t nevent = (Int_t)tree->GetEntries();
@@ -62,68 +47,46 @@ static void BM_ATLAS_Decompress(benchmark::State &state, int algo) {
         for (ev = 0; ev < nevent; ev++) {
             nb += tree->GetEntry(ev);
         }
+
+        hfile->Close();
+
     }
-}
 
-static void BM_ATLAS_Compress_ZLIB(benchmark::State &state) {
-    BM_ATLAS_Compress(state, 1);
-}
-static void BM_ATLAS_Compress_LZMA(benchmark::State &state) {
-    BM_ATLAS_Compress(state, 2);
-}
-static void BM_ATLAS_Compress_LZ4(benchmark::State &state) {
-    BM_ATLAS_Compress(state, 4);
-}
-static void BM_ATLAS_Compress_ZSTD(benchmark::State &state) {
-    BM_ATLAS_Compress(state, 5);
+    gSystem->Exec(("rm -f " + new_filename).c_str());
 }
 
-static void BM_ATLAS_Decompress_ZLIB(benchmark::State &state) {
-    BM_ATLAS_Decompress(state, 1);
+// Benchmarks for ATLAS files of experimental (real) data are deleted
+// because it takes very long time.
+// TODO: find smaller file
+
+static void BM_ATLAS_Decompress_MC_ZLIB(benchmark::State &state) {
+    BM_ATLAS_Decompress(state, 1, "atlasopendata_mc_117050.ttbar_lep");
 }
-static void BM_ATLAS_Decompress_LZMA(benchmark::State &state) {
-    BM_ATLAS_Decompress(state, 2);
+static void BM_ATLAS_Decompress_MC_LZMA(benchmark::State &state) {
+    BM_ATLAS_Decompress(state, 2, "atlasopendata_mc_117050.ttbar_lep");
 }
-static void BM_ATLAS_Decompress_LZ4(benchmark::State &state) {
-    BM_ATLAS_Decompress(state, 4);
+static void BM_ATLAS_Decompress_MC_LZ4(benchmark::State &state) {
+    BM_ATLAS_Decompress(state, 4, "atlasopendata_mc_117050.ttbar_lep");
 }
-static void BM_ATLAS_Decompress_ZSTD(benchmark::State &state) {
-    BM_ATLAS_Decompress(state, 5);
+static void BM_ATLAS_Decompress_MC_ZSTD(benchmark::State &state) {
+    BM_ATLAS_Decompress(state, 5, "atlasopendata_mc_117050.ttbar_lep");
 }
 
-
-BENCHMARK(BM_ATLAS_Compress_ZLIB)
-->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
-
-BENCHMARK(BM_ATLAS_Compress_LZMA)
-->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
-
-BENCHMARK(BM_ATLAS_Compress_LZ4)
-->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
-
-BENCHMARK(BM_ATLAS_Compress_ZSTD)
-->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
-
-
-BENCHMARK(BM_ATLAS_Decompress_ZLIB)
+BENCHMARK(BM_ATLAS_Decompress_MC_ZLIB)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(2);
 
-BENCHMARK(BM_ATLAS_Decompress_LZMA)
+BENCHMARK(BM_ATLAS_Decompress_MC_LZMA)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(2);
 
-BENCHMARK(BM_ATLAS_Decompress_LZ4)
+BENCHMARK(BM_ATLAS_Decompress_MC_LZ4)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(2);
 
-BENCHMARK(BM_ATLAS_Decompress_ZSTD)
+BENCHMARK(BM_ATLAS_Decompress_MC_ZSTD)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(2);
 
 
-BENCHMARK_MAIN(); 
+BENCHMARK_MAIN();

root/io/io/TFile_LHCb_Benchmarks.cxx

@@ -1,5 +1,6 @@
 #include "TFile.h"
 #include "TTree.h"
+#include "TSystem.h"
 
 #include "benchmark/benchmark.h"
 #include "rootbench/RBConfig.h"
@@ -20,42 +21,22 @@ static std::string GetAlgoName(int algo) {
         return "error";
 }
 
-static void BM_LHCb_Compress(benchmark::State &state, int algo) {
-    TFile *oldfile = new TFile((RB::GetDataDir() + "/lhcb_B2ppKK2011_md_noPIDstrip.root").c_str());
-    TTree *oldtree1 = (TTree*)oldfile->Get("TupleB2ppKK/DecayTree");
-    TTree *oldtree2 = (TTree*)oldfile->Get("TupleB2ppKPi/DecayTree");
-    TTree *oldtree3 = (TTree*)oldfile->Get("TupleB2ppPiPi/DecayTree");
+static void BM_LHCb_Decompress(benchmark::State &state, int algo) {
 
     int comp_level = state.range(0);
-    std::string filename = "level_" + std::to_string(comp_level) + "_lhcb_" + GetAlgoName(algo) + ".root";
-
-    for (auto _ : state) {
-        state.PauseTiming();
-
-        TFile *newfile = new TFile(filename.c_str(), "recreate");
-        TTree *newtree1 = oldtree1->CloneTree();
-        TTree *newtree2 = oldtree2->CloneTree();
-        TTree *newtree3 = oldtree3->CloneTree();
-        newfile->SetCompressionAlgorithm(algo);
-        newfile->SetCompressionLevel(comp_level);
+    std::string comp_setting = std::to_string(algo * 100 + comp_level);
+    std::string old_filename = (RB::GetDataDir() + "/lhcb_B2ppKK2011_md_noPIDstrip.root").c_str();
+    std::string new_filename = (RB::GetTempFs() + "level_" + std::to_string(comp_level) + "_lhcb_" + GetAlgoName(algo) + ".root");
 
-        state.ResumeTiming();
-        newfile->Write();
-        state.PauseTiming();
+    gSystem->Exec(("hadd -v 0 -f" + comp_setting + " " + new_filename + " " + old_filename).c_str());
 
-        state.counters["comp_size"] = newfile->GetBytesWritten();
-        newfile->Close();
-
-        state.ResumeTiming();
-    }
-}
+    TFile *newfile = new TFile(new_filename.c_str());
+    state.counters["comp_size"] = newfile->GetSize();
+    newfile->Close();
 
-static void BM_LHCb_Decompress(benchmark::State &state, int algo) {
-    int comp_level = state.range(0);
-
-    std::string filename = "level_" + std::to_string(comp_level) + "_lhcb_" + GetAlgoName(algo) + ".root";
     for (auto _ : state) {
-        TFile *hfile = new TFile(filename.c_str());
+
+        TFile *hfile = new TFile(new_filename.c_str());
         TTree *tree1 = (TTree*)hfile->Get("TupleB2ppKK/DecayTree");
         TTree *tree2 = (TTree*)hfile->Get("TupleB2ppKPi/DecayTree");
         TTree *tree3 = (TTree*)hfile->Get("TupleB2ppPiPi/DecayTree");
@@ -78,22 +59,15 @@ static void BM_LHCb_Decompress(benchmark::State &state, int algo) {
         for (ev = 0; ev < nevent3; ev++) {
             nb += tree3->GetEntry(ev);
         }
+
+        hfile->Close();
+
     }
-}
 
-static void BM_LHCb_Compress_ZLIB(benchmark::State &state) {
-    BM_LHCb_Compress(state, 1);
-}
-static void BM_LHCb_Compress_LZMA(benchmark::State &state) {
-    BM_LHCb_Compress(state, 2);
-}
-static void BM_LHCb_Compress_LZ4(benchmark::State &state) {
-    BM_LHCb_Compress(state, 4);
-}
-static void BM_LHCb_Compress_ZSTD(benchmark::State &state) {
-    BM_LHCb_Compress(state, 5);
+    gSystem->Exec(("rm -f " + new_filename).c_str());
 }
 
+
 static void BM_LHCb_Decompress_ZLIB(benchmark::State &state) {
     BM_LHCb_Decompress(state, 1);
 }
@@ -108,38 +82,21 @@ static void BM_LHCb_Decompress_ZSTD(benchmark::State &state) {
 }
 
 
-BENCHMARK(BM_LHCb_Compress_ZLIB)
-->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
-
-BENCHMARK(BM_LHCb_Compress_LZMA)
-->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
-
-BENCHMARK(BM_LHCb_Compress_LZ4)
-->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
-
-BENCHMARK(BM_LHCb_Compress_ZSTD)
-->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
-
-
 BENCHMARK(BM_LHCb_Decompress_ZLIB)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(3);
 
 BENCHMARK(BM_LHCb_Decompress_LZMA)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(3);
 
 BENCHMARK(BM_LHCb_Decompress_LZ4)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(3);
 
 BENCHMARK(BM_LHCb_Decompress_ZSTD)
 ->Arg(1)->Arg(6)->Arg(9)
-->Unit(benchmark::kMillisecond)->Iterations(5);
+->Unit(benchmark::kMillisecond)->Iterations(3);
 
 
-BENCHMARK_MAIN(); 
+BENCHMARK_MAIN();