added a script to get the latest regression results

cmake/tbb.cmake

@@ -9,18 +9,17 @@ find_package(Git REQUIRED)
 ExternalProject_Add(
     tbb_src
     PREFIX "vendor/intel/tbb"
-    GIT_REPOSITORY "https://github.com/wjakob/tbb.git"
-    GIT_TAG b066defc0229a1e92d7a200eb3fe0f7e35945d95
+    GIT_REPOSITORY "https://github.com/seb711/oneTBB.git"
+    GIT_TAG master
     TIMEOUT 10
     BUILD_COMMAND  make
     UPDATE_COMMAND "" # to prevent rebuilding everytime
     INSTALL_COMMAND ""
     CMAKE_ARGS
-    -DCMAKE_INSTALL_PREFIX=${CMAKE_BINARY_DIR}/vendor/tbb_cpp
     -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
     -DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}
-    -DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS}
-    -DCMAKE_BUILD_TYPE:STRING=${CMAKE_BUILD_TYPE}
+    -DCMAKE_LIBRARY_OUTPUT_DIRECTORY:STRING=./lib
+    -DCMAKE_CXX_FLAGS=${CMAKE_CXX_FLAGS} -DTBB_TEST:BOOL=OFF
 )
 
 # Prepare json

tools/conversion/btrtocsv.cpp

@@ -13,7 +13,8 @@
 #include "yaml-cpp/yaml.h"
 #include "spdlog/spdlog.h"
 #include "tbb/parallel_for.h"
-#include "tbb/task_scheduler_init.h"
+#define TBB_PREVIEW_GLOBAL_CONTROL 1
+#include "tbb/global_control.h"
 // ------------------------------------------------------------------------------
 // Btrfiles library
 #include "btrfiles.hpp"
@@ -106,7 +107,7 @@ int main(int argc, char **argv)
     SchemePool::refresh();
 
     // Init TBB TODO: is that actually still necessary ?
-    tbb::task_scheduler_init init(FLAGS_threads); // NOLINT(cppcoreguidelines-narrowing-conversions)
+    tbb::global_control c(tbb::global_control::max_allowed_parallelism, FLAGS_threads);
 
     // Open output file
     auto csvstream = std::ofstream(FLAGS_csv);

tools/conversion/csvtobtr.cpp

@@ -15,7 +15,9 @@
 #include <yaml-cpp/yaml.h>
 #include <spdlog/spdlog.h>
 #include <tbb/parallel_for.h>
-#include <tbb/task_scheduler_init.h>
+#include <thread>
+#define TBB_PREVIEW_GLOBAL_CONTROL 1
+#include "tbb/global_control.h"
 // ------------------------------------------------------------------------------
 // Btr internal includes
 #include "common/Utils.hpp"
@@ -42,7 +44,7 @@ DEFINE_bool(create_btr, false, "If false will exit after binary creation");
 DEFINE_bool(verify, true, "Verify that decompression works");
 DEFINE_int32(chunk, -1, "Select a specific chunk to measure");
 DEFINE_int32(column, -1, "Select a specific column to measure");
-DEFINE_uint32(threads, 8, "");
+DEFINE_uint32(threads, -1, "");
 // ------------------------------------------------------------------------------
 using namespace btrblocks;
 // ------------------------------------------------------------------------------
@@ -72,8 +74,13 @@ int main(int argc, char **argv)
     // This seems necessary to be
     SchemePool::refresh();
 
-    // Init TBB TODO: is that actually still necessary ?
-    tbb::task_scheduler_init init(FLAGS_threads);
+    if (FLAGS_threads < 1) {
+      tbb::global_control c(tbb::global_control::max_allowed_parallelism,
+                            std::thread::hardware_concurrency());
+    } else {
+      tbb::global_control c(tbb::global_control::max_allowed_parallelism,
+                            FLAGS_threads);
+    }
 
     // Load schema
     const auto schema = YAML::LoadFile(FLAGS_yaml);
@@ -124,7 +131,7 @@ int main(int argc, char **argv)
 
     // Prepare datastructures for btr compression
     //auto ranges = relation.getRanges(static_cast<SplitStrategy>(1), 9999);
-    auto ranges = relation.getRanges(SplitStrategy::SEQUENTIAL, 9999);
+    auto ranges = relation.getRanges(SplitStrategy::SEQUENTIAL, -1);
     assert(ranges.size() > 0);
     Datablock datablockV2(relation);
     std::filesystem::create_directory(FLAGS_btr);

tools/conversion/decompression-speed.cpp

@@ -6,7 +6,8 @@
 // -------------------------------------------------------------------------------------
 #include "gflags/gflags.h"
 #include "tbb/parallel_for.h"
-#include "tbb/task_scheduler_init.h"
+#define TBB_PREVIEW_GLOBAL_CONTROL 1
+#include "tbb/global_control.h"
 // -------------------------------------------------------------------------------------
 #include "common/PerfEvent.hpp"
 #include "common/Utils.hpp"
@@ -108,7 +109,7 @@ int main(int argc, char **argv) {
     } else {
         threads = FLAGS_threads;
     }
-    tbb::task_scheduler_init init(threads);
+    tbb::global_control c(tbb::global_control::max_allowed_parallelism, threads);
 
     // Read the metadata
     std::vector<char> raw_file_metadata;

tools/playground/generate_s3_data.cpp

@@ -6,7 +6,8 @@
 #include <vector>
 // -------------------------------------------------------------------------------------
 #include <tbb/parallel_for.h>
-#include <tbb/task_scheduler_init.h>
+#define TBB_PREVIEW_GLOBAL_CONTROL 1
+#include <tbb/global_control.h>
 // -------------------------------------------------------------------------------------
 #include <aws/core/Aws.h>
 #include <aws/s3-crt/S3CrtClient.h>
@@ -180,9 +181,7 @@ int main(int argc, char** argv) {
   std::stringstream bucket;
   bucket << bucket_prefix << "-" << region;
 
-  // tbb::task_scheduler_init init(1);
-
-  // tbb::task_scheduler_init init(1);
+  //  tbb::global_control c(tbb::global_control::max_allowed_parallelism, 1);
 
   Aws::SDKOptions options;
   Aws::InitAPI(options);

tools/regression-benchmark/benchmark.sh

@@ -0,0 +1,61 @@
+#!/usr/bin/env bash
+
+# Exit on error
+set -e
+
+# Function to read CSV file and sync URIs
+sync_uris() {
+  input_file="$1"
+  output_file="$2"
+  index=1
+
+  while IFS=',' read -r name uri yaml; do
+    schemaname=$(basename "$yaml")
+
+    echo $uri
+
+    if [[ ! -f "./csvtobtrdata/yaml/$name/$schemaname" ]]; then
+      mkdir ./csvtobtrdata/yaml/$name -p
+      aws s3 cp $yaml ./csvtobtrdata/yaml/$name/ --request-payer requester
+    fi
+
+    btr_dir="./csvtobtrdata/btrblocks/$name/"
+    mkdir -p "$btr_dir" || rm -rf "${$btr_dir:?}"/*
+    bin_dir="./csvtobtrdata/btrblocks_bin/$name/"
+    echo "aws s3 sync $uri $bin_dir --request-payer requester"
+    if [[ ! -d $bin_dir ]]; then
+      aws s3 sync --request-payer requester $uri $bin_dir
+    fi
+
+    yaml_file="./csvtobtrdata/yaml/$name/$schemaname"
+    ./csvtobtr --btr $btr_dir --binary $bin_dir --create_btr true --yaml $yaml_file
+
+    echo "$name, $(./decompression-speed --btr $btr_dir --reps 100 --binary $bin_dir --yaml $yaml_file --verify)" >> $output_file
+
+    ((index++))
+
+  done < "$input_file"
+}
+
+# install things
+# sudo apt-get update && sudo apt-get install libssl-dev libcurl4-openssl-dev -y
+command -v aws &>/dev/null || { echo >&2 "Please install the aws cli"; exit 1; }
+
+# build the benchmark thing
+output_file="results.csv"
+rm -f $output_file
+mkdir -p tmpbuild
+cd tmpbuild
+
+dataset="../datasets.csv"
+# Check if uris.csv exists
+if [[ ! -f $dataset ]]; then
+  echo "$dataset file not found."
+  exit 1
+fi
+
+cmake ../../.. -DCMAKE_BUILD_TYPE=Release
+make -j csvtobtr
+make -j decompression-speed
+sync_uris $dataset $output_file
+

tools/regression-benchmark/datasets.csv

@@ -0,0 +1,43 @@
+Arade_1, s3://public-bi-eu-central-1/v0.0.1/binary/Arade_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Arade_1/Arade_1.yaml
+Bimbo_1, s3://public-bi-eu-central-1/v0.0.1/binary/Bimbo_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Bimbo_1/Bimbo_1.yaml
+CityMaxCapita_1, s3://public-bi-eu-central-1/v0.0.1/binary/CityMaxCapita_1/, s3://public-bi-eu-central-1/v0.0.1/binary/CityMaxCapita_1/CityMaxCapita_1.yaml
+CMSprovider_1, s3://public-bi-eu-central-1/v0.0.1/binary/CMSprovider_1/, s3://public-bi-eu-central-1/v0.0.1/binary/CMSprovider_1/CMSprovider_1.yaml
+CommonGovernment_1, s3://public-bi-eu-central-1/v0.0.1/binary/CommonGovernment_1/, s3://public-bi-eu-central-1/v0.0.1/binary/CommonGovernment_1/CommonGovernment_1.yaml
+Cooperations_1, s3://public-bi-eu-central-1/v0.0.1/binary/Corporations_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Corporations_1/Corporations_1.yaml
+Eixo_1, s3://public-bi-eu-central-1/v0.0.1/binary/Eixo_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Eixo_1/Eixo_1.yaml
+Euro2016_1, s3://public-bi-eu-central-1/v0.0.1/binary/Euro2016_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Euro2016_1/Euro2016_1.yaml
+Food_1, s3://public-bi-eu-central-1/v0.0.1/binary/Food_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Food_1/Food_1.yaml
+Generico_1, s3://public-bi-eu-central-1/v0.0.1/binary/Generico_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Generico_1/Generico_1.yaml
+Hashtags_1, s3://public-bi-eu-central-1/v0.0.1/binary/HashTags_1/, s3://public-bi-eu-central-1/v0.0.1/binary/HashTags_1/HashTags_1.yaml
+IGlocations_2, s3://public-bi-eu-central-1/v0.0.1/binary/IGlocations2_1/, s3://public-bi-eu-central-1/v0.0.1/binary/IGlocations2_1/IGlocations2_1.yaml
+Medicare1_1, s3://public-bi-eu-central-1/v0.0.1/binary/Medicare1_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Medicare1_1/Medicare1_1.yaml
+Medicare2_1, s3://public-bi-eu-central-1/v0.0.1/binary/Medicare2_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Medicare2_1/Medicare2_1.yaml
+Medicare3_1, s3://public-bi-eu-central-1/v0.0.1/binary/Medicare3_1/,  s3://public-bi-eu-central-1/v0.0.1/binary/Medicare3_1/Medicare3_1.yaml
+MedPayment1_1, s3://public-bi-eu-central-1/v0.0.1/binary/MedPayment1_1/,  s3://public-bi-eu-central-1/v0.0.1/binary/MedPayment1_1/MedPayment1_1.yaml
+MedPayment2_1, s3://public-bi-eu-central-1/v0.0.1/binary/MedPayment2_1/,  s3://public-bi-eu-central-1/v0.0.1/binary/MedPayment2_1/MedPayment2_1.yaml
+MLB_68, s3://public-bi-eu-central-1/v0.0.1/binary/68/MLB_68/, s3://public-bi-eu-central-1/v0.0.1/binary/MLB/68/MLB_68.yaml
+Motos_1, s3://public-bi-eu-central-1/v0.0.1/binary/Motos_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Motos_1/Motos_1.yaml
+MulheresMil_1, s3://public-bi-eu-central-1/v0.0.1/binary/MulheresMil_1/, s3://public-bi-eu-central-1/v0.0.1/binary/MulheresMil_1/MulheresMil_1.yaml
+NYC_1, s3://public-bi-eu-central-1/v0.0.1/binary/NYC_1/, s3://public-bi-eu-central-1/v0.0.1/binary/NYC_1/NYC_1.yaml
+PanCreactomy1_1, s3://public-bi-eu-central-1/v0.0.1/binary/PanCreactomy1_1/, s3://public-bi-eu-central-1/v0.0.1/binary/PanCreactomy1_1/PanCreactomy1_1.yaml
+PanCreactomy2_1, s3://public-bi-eu-central-1/v0.0.1/binary/PanCreactomy2_1/, s3://public-bi-eu-central-1/v0.0.1/binary/PanCreactomy2_1/PanCreactomy2_1.yaml
+Physicians_1, s3://public-bi-eu-central-1/v0.0.1/binary/Physicians_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Physicians_1/Physicians_1.yaml
+Provider_1, s3://public-bi-eu-central-1/v0.0.1/binary/Provider_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Provider_1/Provider_1.yaml
+RealEstate1_1, s3://public-bi-eu-central-1/v0.0.1/binary/RealEstate1_1/, s3://public-bi-eu-central-1/v0.0.1/binary/RealEstate1_1/RealEstate1_1.yaml
+RealEstate2_1, s3://public-bi-eu-central-1/v0.0.1/binary/RealEstate2_1/,  s3://public-bi-eu-central-1/v0.0.1/binary/RealEstate2_1/RealEstate2_1.yaml
+Redfin1_1, s3://public-bi-eu-central-1/v0.0.1/binary/Redfin1_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Redfin1_1/Redfin1_1.yaml
+Redfin2_1, s3://public-bi-eu-central-1/v0.0.1/binary/Redfin2_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Redfin2_1/Redfin2_1.yaml
+Redfin3_1, s3://public-bi-eu-central-1/v0.0.1/binary/Redfin3_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Redfin3_1/Redfin3_1.yaml
+Redfin4_1, s3://public-bi-eu-central-1/v0.0.1/binary/Redfin4_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Redfin4_1/Redfin4_1.yaml
+Rentabilidad_1, s3://public-bi-eu-central-1/v0.0.1/binary/Rentabilidad_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Rentabilidad_1/Rentabilidad_1.yaml
+Romance_1, s3://public-bi-eu-central-1/v0.0.1/binary/Romance_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Romance_1/Romance_1.yaml
+SalariesFrance_1, s3://public-bi-eu-central-1/v0.0.1/binary/SalariesFrance_1/, s3://public-bi-eu-central-1/v0.0.1/binary/SalariesFrance_1/SalariesFrance_1.yaml
+TableroSistemaPenal_1, s3://public-bi-eu-central-1/v0.0.1/binary/TableroSistemaPenal_1/, s3://public-bi-eu-central-1/v0.0.1/binary/TableroSistemaPenal_1/TableroSistemaPenal_1.yaml
+Taxpayer_1, s3://public-bi-eu-central-1/v0.0.1/binary/Taxpayer_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Taxpayer_1/Taxpayer_1.yaml
+Telco_1, s3://public-bi-eu-central-1/v0.0.1/binary/Telco_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Telco_1/Telco_1.yaml
+TrainsUK1_4, s3://public-bi-eu-central-1/v0.0.1/binary/4/TrainsUK1_4/, s3://public-bi-eu-central-1/v0.0.1/binary/TrainsUK1/4/TrainsUK1_4.yaml
+TrainsUK2_1, s3://public-bi-eu-central-1/v0.0.1/binary/TrainsUK2_1/,  s3://public-bi-eu-central-1/v0.0.1/binary/TrainsUK2_1/TrainsUK2_1.yaml
+Uberlandia_1, s3://public-bi-eu-central-1/v0.0.1/binary/Uberlandia_1/,  s3://public-bi-eu-central-1/v0.0.1/binary/Uberlandia_1/Uberlandia_1.yaml
+USCensus_1, s3://public-bi-eu-central-1/v0.0.1/binary/USCensus_1/,  s3://public-bi-eu-central-1/v0.0.1/binary/USCensus_1/USCensus_1.yaml
+Wins_1, s3://public-bi-eu-central-1/v0.0.1/binary/Wins_1/, s3://public-bi-eu-central-1/v0.0.1/binary/Wins_1/Wins_1.yaml
+YaleLanguages_1, s3://public-bi-eu-central-1/v0.0.1/binary/YaleLanguages_1/,  s3://public-bi-eu-central-1/v0.0.1/binary/YaleLanguages_1/YaleLanguages_1.yaml