Release patch release 1.1.1 with faster CPU performance (#5732)

* Fix release degradation (#5720)

* fix release degradation, related to 5666

* less resizes

Co-authored-by: SHVETS, KIRILL <kirill.shvets@intel.com>

* Make 1.1.1 patch release

* Disable too-many-function-args pylint warning for predict()

* Fix Windows CI

* Remove cpplint

Co-authored-by: ShvetsKS <33296480+ShvetsKS@users.noreply.github.com>
Co-authored-by: SHVETS, KIRILL <kirill.shvets@intel.com>

CMakeLists.txt

@@ -1,5 +1,5 @@
 cmake_minimum_required(VERSION 3.13)
-project(xgboost LANGUAGES CXX C VERSION 1.1.0)
+project(xgboost LANGUAGES CXX C VERSION 1.1.1)
 include(cmake/Utils.cmake)
 list(APPEND CMAKE_MODULE_PATH "${xgboost_SOURCE_DIR}/cmake/modules")
 cmake_policy(SET CMP0022 NEW)
@@ -57,6 +57,7 @@ address, leak, undefined and thread.")
 ## Plugins
 option(PLUGIN_LZ4 "Build lz4 plugin" OFF)
 option(PLUGIN_DENSE_PARSER "Build dense parser plugin" OFF)
+option(ADD_PKGCONFIG "Add xgboost.pc into system." ON)
 
 #-- Checks for building XGBoost
 if (USE_DEBUG_OUTPUT AND (NOT (CMAKE_BUILD_TYPE MATCHES Debug)))
@@ -274,3 +275,12 @@ endif (GOOGLE_TEST)
 # replace /MD with /MT. See https://github.com/dmlc/xgboost/issues/4462
 # for issues caused by mixing of /MD and /MT flags
 msvc_use_static_runtime()
+
+# Add xgboost.pc
+if (ADD_PKGCONFIG)
+  configure_file(${xgboost_SOURCE_DIR}/cmake/xgboost.pc.in ${xgboost_BINARY_DIR}/xgboost.pc @ONLY)
+
+  install(
+    FILES ${xgboost_BINARY_DIR}/xgboost.pc
+    DESTINATION ${CMAKE_INSTALL_LIBDIR}/pkgconfig)
+endif (ADD_PKGCONFIG)

Jenkinsfile

@@ -48,7 +48,6 @@ pipeline {
         script {
           parallel ([
             'clang-tidy': { ClangTidy() },
-            'lint': { Lint() },
             'sphinx-doc': { SphinxDoc() },
             'doxygen': { Doxygen() }
           ])

Jenkinsfile-win64

@@ -28,7 +28,7 @@ pipeline {
       steps {
         script {
           parallel ([
-            'build-win64-cuda9.0': { BuildWin64() }
+            'build-win64-cuda10.0': { BuildWin64() }
           ])
         }
         milestone ordinal: 2
@@ -40,7 +40,6 @@ pipeline {
         script {
           parallel ([
             'test-win64-cpu': { TestWin64CPU() },
-            'test-win64-gpu-cuda9.0': { TestWin64GPU(cuda_target: 'cuda9') },
             'test-win64-gpu-cuda10.0': { TestWin64GPU(cuda_target: 'cuda10_0') },
             'test-win64-gpu-cuda10.1': { TestWin64GPU(cuda_target: 'cuda10_1') }
           ])
@@ -67,7 +66,7 @@ def checkoutSrcs() {
 }
 
 def BuildWin64() {
-  node('win64 && build') {
+  node('win64 && build && cuda10') {
     unstash name: 'srcs'
     echo "Building XGBoost for Windows AMD64 target..."
     bat "nvcc --version"

R-package/DESCRIPTION

@@ -1,7 +1,7 @@
 Package: xgboost
 Type: Package
 Title: Extreme Gradient Boosting
-Version: 1.1.0.1
+Version: 1.1.1.1
 Date: 2020-02-21
 Authors@R: c(
   person("Tianqi", "Chen", role = c("aut"),

R-package/configure

@@ -2698,7 +2698,7 @@ fi
 if test `uname -s` = "Darwin"
 then
   OPENMP_CXXFLAGS='-Xclang -fopenmp'
-  OPENMP_LIB='/usr/local/lib/libomp.dylib'
+  OPENMP_LIB='-lomp'
   ac_pkg_openmp=no
   { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether OpenMP will work in a package" >&5
 $as_echo_n "checking whether OpenMP will work in a package... " >&6; }

R-package/configure.ac

@@ -29,7 +29,7 @@ fi
 if test `uname -s` = "Darwin"
 then
   OPENMP_CXXFLAGS='-Xclang -fopenmp'
-  OPENMP_LIB='/usr/local/lib/libomp.dylib'
+  OPENMP_LIB='-lomp'
   ac_pkg_openmp=no
   AC_MSG_CHECKING([whether OpenMP will work in a package])
   AC_LANG_CONFTEST([AC_LANG_PROGRAM([[#include <omp.h>]], [[ return (omp_get_max_threads() <= 1); ]])])

cmake/xgboost.pc.in

@@ -0,0 +1,12 @@
+prefix=@CMAKE_INSTALL_PREFIX@
+version=@xgboost_VERSION@
+exec_prefix=${prefix}/bin
+libdir=${prefix}/lib
+includedir=${prefix}/include
+
+Name: xgboost
+Description: XGBoost - Scalable and Flexible Gradient Boosting.
+Version: ${version}
+
+Cflags: -I${includedir}
+Libs: -L${libdir} -lxgboost

doc/tutorials/aft_survival_analysis.rst

@@ -35,7 +35,7 @@ There are four kinds of censoring:
 
 * **Uncensored**: the label is not censored and given as a single number.
 * **Right-censored**: the label is of form :math:`[a, +\infty)`, where :math:`a` is the lower bound.
-* **Left-censored**: the label is of form :math:`(-\infty, b]`, where :math:`b` is the upper bound.
+* **Left-censored**: the label is of form :math:`[0, b]`, where :math:`b` is the upper bound.
 * **Interval-censored**: the label is of form :math:`[a, b]`, where :math:`a` and :math:`b` are the lower and upper bounds, respectively.
 
 Right-censoring is the most commonly used.
@@ -83,7 +83,7 @@ Censoring type    Interval form        Lower bound finite? Upper bound finite?
 ================= ==================== =================== ===================
 Uncensored        :math:`[a, a]`       |tick|              |tick|
 Right-censored    :math:`[a, +\infty)` |tick|              |cross|
-Left-censored     :math:`(-\infty, b]` |cross|             |tick|
+Left-censored     :math:`[0, b]`       |tick|              |tick|
 Interval-censored :math:`[a, b]`       |tick|              |tick|
 ================= ==================== =================== ===================
 
@@ -102,7 +102,7 @@ Collect the lower bound numbers in one array (let's call it ``y_lower_bound``) a
   # Associate ranged labels with the data matrix.
   # This example shows each kind of censored labels.
   #                         uncensored    right     left  interval
-  y_lower_bound = np.array([      2.0,     3.0, -np.inf,     4.0])
+  y_lower_bound = np.array([      2.0,     3.0,     0.0,     4.0])
   y_upper_bound = np.array([      2.0, +np.inf,     4.0,     5.0])
   dtrain.set_float_info('label_lower_bound', y_lower_bound)
   dtrain.set_float_info('label_upper_bound', y_upper_bound)
@@ -120,7 +120,7 @@ Collect the lower bound numbers in one array (let's call it ``y_lower_bound``) a
   # Associate ranged labels with the data matrix.
   # This example shows each kind of censored labels.
   #                   uncensored  right  left  interval
-  y_lower_bound <- c(        2.,    3., -Inf,       4.)
+  y_lower_bound <- c(        2.,    3.,   0.,       4.)
   y_upper_bound <- c(        2.,  +Inf,   4.,       5.)
   setinfo(dtrain, 'label_lower_bound', y_lower_bound)
   setinfo(dtrain, 'label_upper_bound', y_upper_bound)
@@ -136,7 +136,7 @@ Now we are ready to invoke the training API:
             'aft_loss_distribution_scale': 1.20,
             'tree_method': 'hist', 'learning_rate': 0.05, 'max_depth': 2}
   bst = xgb.train(params, dtrain, num_boost_round=5,
-                  evals=[(dtrain, 'train'), (dvalid, 'valid')])
+                  evals=[(dtrain, 'train')])
 
 .. code-block:: r
   :caption: R
@@ -165,4 +165,4 @@ Currently, you can choose from three probability distributions for ``aft_loss_di
 ``extreme``               :math:`e^z e^{-\exp{z}}`
 ========================= ===========================================
 
-Note that it is not yet possible to set the ranged label using the scikit-learn interface (e.g. :class:`xgboost.XGBRegressor`). For now, you should use :class:`xgboost.train` with :class:`xgboost.DMatrix`.
+Note that it is not yet possible to set the ranged label using the scikit-learn interface (e.g. :class:`xgboost.XGBRegressor`). For now, you should use :class:`xgboost.train` with :class:`xgboost.DMatrix`.

jvm-packages/pom.xml

@@ -6,7 +6,7 @@
 
     <groupId>ml.dmlc</groupId>
     <artifactId>xgboost-jvm_2.12</artifactId>
-    <version>1.1.0</version>
+    <version>1.1.1</version>
     <packaging>pom</packaging>
     <name>XGBoost JVM Package</name>
     <description>JVM Package for XGBoost</description>

jvm-packages/xgboost4j-example/pom.xml

@@ -6,10 +6,10 @@
     <parent>
         <groupId>ml.dmlc</groupId>
         <artifactId>xgboost-jvm_2.12</artifactId>
-        <version>1.1.0</version>
+        <version>1.1.1</version>
     </parent>
     <artifactId>xgboost4j-example_2.12</artifactId>
-    <version>1.1.0</version>
+    <version>1.1.1</version>
     <packaging>jar</packaging>
     <build>
         <plugins>
@@ -26,7 +26,7 @@
         <dependency>
             <groupId>ml.dmlc</groupId>
             <artifactId>xgboost4j-spark_${scala.binary.version}</artifactId>
-            <version>1.1.0</version>
+            <version>1.1.1</version>
         </dependency>
         <dependency>
             <groupId>org.apache.spark</groupId>
@@ -37,7 +37,7 @@
         <dependency>
             <groupId>ml.dmlc</groupId>
             <artifactId>xgboost4j-flink_${scala.binary.version}</artifactId>
-            <version>1.1.0</version>
+            <version>1.1.1</version>
         </dependency>
         <dependency>
             <groupId>org.apache.commons</groupId>

jvm-packages/xgboost4j-flink/pom.xml

@@ -6,10 +6,10 @@
     <parent>
         <groupId>ml.dmlc</groupId>
         <artifactId>xgboost-jvm_2.12</artifactId>
-        <version>1.1.0</version>
+        <version>1.1.1</version>
     </parent>
     <artifactId>xgboost4j-flink_2.12</artifactId>
-    <version>1.1.0</version>
+    <version>1.1.1</version>
     <build>
         <plugins>
             <plugin>
@@ -26,7 +26,7 @@
         <dependency>
             <groupId>ml.dmlc</groupId>
             <artifactId>xgboost4j_${scala.binary.version}</artifactId>
-            <version>1.1.0</version>
+            <version>1.1.1</version>
         </dependency>
         <dependency>
             <groupId>org.apache.commons</groupId>

jvm-packages/xgboost4j-spark/pom.xml

@@ -6,7 +6,7 @@
     <parent>
         <groupId>ml.dmlc</groupId>
         <artifactId>xgboost-jvm_2.12</artifactId>
-        <version>1.1.0</version>
+        <version>1.1.1</version>
     </parent>
     <artifactId>xgboost4j-spark_2.12</artifactId>
     <build>
@@ -24,7 +24,7 @@
         <dependency>
             <groupId>ml.dmlc</groupId>
             <artifactId>xgboost4j_${scala.binary.version}</artifactId>
-            <version>1.1.0</version>
+            <version>1.1.1</version>
         </dependency>
         <dependency>
             <groupId>org.apache.spark</groupId>

jvm-packages/xgboost4j/pom.xml

@@ -6,10 +6,10 @@
     <parent>
         <groupId>ml.dmlc</groupId>
         <artifactId>xgboost-jvm_2.12</artifactId>
-        <version>1.1.0</version>
+        <version>1.1.1</version>
     </parent>
     <artifactId>xgboost4j_2.12</artifactId>
-    <version>1.1.0</version>
+    <version>1.1.1</version>
     <packaging>jar</packaging>
 
     <dependencies>

python-package/xgboost/VERSION

@@ -1 +1 @@
-1.1.0
+1.1.1

python-package/xgboost/core.py

@@ -1467,6 +1467,7 @@ class Booster(object):
         self._validate_features(data)
         return self.eval_set([(data, name)], iteration)
 
+    # pylint: disable=too-many-function-args
     def predict(self,
                 data,
                 output_margin=False,
@@ -1588,7 +1589,7 @@ class Booster(object):
                 ngroup = int(chunk_size / ((data.num_col() + 1) *
                                            (data.num_col() + 1)))
                 if ngroup == 1:
-                    preds = preds.reshape(nrow,
+                    preds = preds.reshape(nrow,  # pylint: disable=too-many-function-args
                                           data.num_col() + 1,
                                           data.num_col() + 1)
                 else:

src/learner.cc

@@ -689,15 +689,23 @@ class LearnerIO : public LearnerConfiguration {
       warn_old_model = false;
     }
 
-    if (mparam_.major_version >= 1) {
+    if (mparam_.major_version < 1) {
-      learner_model_param_ = LearnerModelParam(mparam_,
-                                               obj_->ProbToMargin(mparam_.base_score));
-    } else {
       // Before 1.0.0, base_score is saved as a transformed value, and there's no version
-      // attribute in the saved model.
+      // attribute (saved a 0) in the saved model.
-      learner_model_param_ = LearnerModelParam(mparam_, mparam_.base_score);
+      std::string multi{"multi:"};
+      if (!std::equal(multi.cbegin(), multi.cend(), tparam_.objective.cbegin())) {
+        HostDeviceVector<float> t;
+        t.HostVector().resize(1);
+        t.HostVector().at(0) = mparam_.base_score;
+        this->obj_->PredTransform(&t);
+        auto base_score = t.HostVector().at(0);
+        mparam_.base_score = base_score;
+      }
       warn_old_model = true;
     }
+
+    learner_model_param_ =
+        LearnerModelParam(mparam_, obj_->ProbToMargin(mparam_.base_score));
     if (attributes_.find("objective") != attributes_.cend()) {
       auto obj_str = attributes_.at("objective");
       auto j_obj = Json::Load({obj_str.c_str(), obj_str.size()});

src/objective/regression_obj.cu

@@ -41,10 +41,11 @@ struct RegLossParam : public XGBoostParameter<RegLossParam> {
 template<typename Loss>
 class RegLossObj : public ObjFunction {
  protected:
-  HostDeviceVector<int> label_correct_;
+  HostDeviceVector<float> additional_input_;
 
  public:
-  RegLossObj() = default;
+  // 0 - label_correct flag, 1 - scale_pos_weight, 2 - is_null_weight
+  RegLossObj(): additional_input_(3) {}
 
   void Configure(const std::vector<std::pair<std::string, std::string> >& args) override {
     param_.UpdateAllowUnknown(args);
@@ -64,8 +65,7 @@ class RegLossObj : public ObjFunction {
     size_t const ndata = preds.Size();
     out_gpair->Resize(ndata);
     auto device = tparam_->gpu_id;
-    label_correct_.Resize(1);
+    additional_input_.HostVector().begin()[0] = 1;  // Fill the label_correct flag
-    label_correct_.Fill(1);
 
     bool is_null_weight = info.weights_.Size() == 0;
     if (!is_null_weight) {
@@ -73,35 +73,37 @@ class RegLossObj : public ObjFunction {
           << "Number of weights should be equal to number of data points.";
     }
     auto scale_pos_weight = param_.scale_pos_weight;
-    common::Transform<>::Init(
+    additional_input_.HostVector().begin()[1] = scale_pos_weight;
-        [=] XGBOOST_DEVICE(size_t _idx,
+    additional_input_.HostVector().begin()[2] = is_null_weight;
-                           common::Span<int> _label_correct,
+
+    common::Transform<>::Init([] XGBOOST_DEVICE(size_t _idx,
+                           common::Span<float> _additional_input,
                            common::Span<GradientPair> _out_gpair,
                            common::Span<const bst_float> _preds,
                            common::Span<const bst_float> _labels,
                            common::Span<const bst_float> _weights) {
+          const float _scale_pos_weight = _additional_input[1];
+          const bool _is_null_weight = _additional_input[2];
+
           bst_float p = Loss::PredTransform(_preds[_idx]);
-          bst_float w = is_null_weight ? 1.0f : _weights[_idx];
+          bst_float w = _is_null_weight ? 1.0f : _weights[_idx];
           bst_float label = _labels[_idx];
           if (label == 1.0f) {
-            w *= scale_pos_weight;
+            w *= _scale_pos_weight;
           }
           if (!Loss::CheckLabel(label)) {
             // If there is an incorrect label, the host code will know.
-            _label_correct[0] = 0;
+            _additional_input[0] = 0;
           }
           _out_gpair[_idx] = GradientPair(Loss::FirstOrderGradient(p, label) * w,
                                           Loss::SecondOrderGradient(p, label) * w);
         },
         common::Range{0, static_cast<int64_t>(ndata)}, device).Eval(
-            &label_correct_, out_gpair, &preds, &info.labels_, &info.weights_);
+            &additional_input_, out_gpair, &preds, &info.labels_, &info.weights_);
 
-    // copy "label correct" flags back to host
+    auto const flag = additional_input_.HostVector().begin()[0];
-    std::vector<int>& label_correct_h = label_correct_.HostVector();
+    if (flag == 0) {
-    for (auto const flag : label_correct_h) {
+      LOG(FATAL) << Loss::LabelErrorMsg();
-      if (flag == 0) {
-        LOG(FATAL) << Loss::LabelErrorMsg();
-      }
     }
   }
 

tests/pytest.ini

@@ -1,3 +1,4 @@
 [pytest]
 markers =
-    mgpu: Mark a test that requires multiple GPUs to run.
+    mgpu: Mark a test that requires multiple GPUs to run.
+    ci: Mark a test that runs only on CI.

tests/python/test_model_compatibility.py

@@ -4,6 +4,7 @@ import generate_models as gm
 import json
 import zipfile
 import pytest
+import copy
 
 
 def run_model_param_check(config):
@@ -124,6 +125,9 @@ def test_model_compatibility():
         if name.startswith('xgboost-'):
             booster = xgboost.Booster(model_file=path)
             run_booster_check(booster, name)
+            # Do full serialization.
+            booster = copy.copy(booster)
+            run_booster_check(booster, name)
         elif name.startswith('xgboost_scikit'):
             run_scikit_model_check(name, path)
         else: