* Initial performance optimizations for xgboost
* remove includes
* revert float->double
* fix for CI
* fix for CI
* fix for CI
* fix for CI
* fix for CI
* fix for CI
* fix for CI
* fix for CI
* fix for CI
* fix for CI
* Check existence of _mm_prefetch and __builtin_prefetch
* Fix lint
* optimizations for CPU
* appling comments in review
* add some comments, code refactoring
* fixing issues in CI
* adding runtime checks
* remove 1 extra check
* remove extra checks in BuildHist
* remove checks
* add debug info
* added debug info
* revert changes
* added comments
* Apply suggestions from code review
Co-Authored-By: Philip Hyunsu Cho <chohyu01@cs.washington.edu>
* apply review comments
* Remove unused function CreateNewNodes()
* Add descriptive comment on node_idx variable in QuantileHistMaker::Builder::BuildHistsBatch()
* Implement tree model dump with a code generator.
* Split up generators.
* Implement graphviz generator.
* Use pattern matching.
* [Breaking] Return a Source in `to_graphviz` instead of Digraph in Python package.
Co-Authored-By: Philip Hyunsu Cho <chohyu01@cs.washington.edu>
* - do not create device vectors for the entire sparse page while computing histograms...
- while creating the compressed histogram indices, the row vector is created for the entire
sparse page batch. this is needless as we only process chunks at a time based on a slice
of the total gpu memory
- this pr will allocate only as much as required to store the ppropriate row indices and the entries
* - do not dereference row_ptrs once the device_vector has been created to elide host copies of those counts
- instead, grab the entry counts directly from the sparsepage
* - set the appropriate device before freeing device memory...
- pr #4532 added a global memory tracker/logger to keep track of number of (de)allocations
and peak memory usage on a per device basis.
- this pr adds the appropriate check to make sure that the (de)allocation counts and memory usages
makes sense for the device. since verbosity is typically increased on debug/non-retail builds.
* - pre-create cub allocators and reuse them
- create them once and not resize them dynamically. we need to ensure that these allocators
are created and destroyed exactly once so that the appropriate device id's are set
This is part 1 of refactoring configuration.
* Move tree heuristic configurations.
* Split up declarations and definitions for GBTree.
* Implement UseGPU in gbm.
* - training with external memory - part 2 of 2
- when external memory support is enabled, building of histogram indices are
done incrementally for every sparse page
- the entire set of input data is divided across multiple gpu's and the relative
row positions within each device is tracked when building the compressed histogram buffer
- this was tested using a mortgage dataset containing ~ 670m rows before 4xt4's could be
saturated
* Fix C++11 config parser
* Use raw strings to improve readability of regex
* Fix compilation for GCC 5.x
Co-authored-by: Jiaming Yuan <jm.yuan@outlook.com>
* - training with external memory part 1 of 2
- this pr focuses on computing the quantiles using multiple gpus on a
dataset that uses the external cache capabilities
- there will a follow-up pr soon after this that will support creation
of histogram indices on large dataset as well
- both of these changes are required to support training with external memory
- the sparse pages in dmatrix are taken in batches and the the cut matrices
are incrementally built
- also snuck in some (perf) changes related to sketches aggregation amongst multiple
features across multiple sparse page batches. instead of aggregating the summary
inside each device and merged later, it is aggregated in-place when the device
is working on different rows but the same feature
* Only define `gpu_id` and `n_gpus` in `LearnerTrainParam`
* Pass LearnerTrainParam through XGBoost vid factory method.
* Disable all GPU usage when GPU related parameters are not specified (fixes XGBoost choosing GPU over aggressively).
* Test learner train param io.
* Fix gpu pickling.
* - fix issues with training with external memory on cpu
- use the batch size to determine the correct number of rows in a batch
- use the right number of threads in omp parallalization if the batch size
is less than the default omp max threads (applicable for the last batch)
* - handle scenarios where last batch size is < available number of threads
- augment tests such that we can test all scenarios (batch size <, >, = number of threads)
* adding support for matrix slicing with query ID for cross-validation
* hail mary test of unrar installation for windows tests
* trying to modify tests to run in Github CI
* Remove dependency on wget and unrar
* Save error log from R test
* Relax assertion in test_training
* Use int instead of bool in C function interface
* Revise R interface
* Add XGDMatrixSliceDMatrixEx and keep old XGDMatrixSliceDMatrix for API compatibility
* Add CMake option to use bundled gtest from dmlc-core, so that it is easy to build XGBoost with gtest on Windows
* Consistently apply OpenMP flag to all targets. Force enable OpenMP when USE_CUDA is turned on.
* Insert vcomp140.dll into Windows wheels
* Add C++ and Python tests for CPU and GPU targets (CUDA 9.0, 10.0, 10.1)
* Prevent spurious msbuild failure
* Add GPU tests
* Upgrade dmlc-core
* Make CMakeLists.txt compatible with CMake 3.3; require CMake 3.11 for MSVC
* Use CMake 3.12 when sanitizer is enabled
* Disable funroll-loops for MSVC
* Use cmake version in container name
* Add missing arg
* Fix egrep use in ci_build.sh
* Display CMake version
* Do not set OpenMP_CXX_LIBRARIES for MSVC
* Use cmake_minimum_required()
* Combine thread launches into single launch per tree for gpu_hist
algorithm.
* Address deprecation warning
* Add manual column sampler constructor
* Turn off omp dynamic to get a guaranteed number of threads
* Enable openmp in cuda code
* All Linux tests are now in Jenkins CI
* Tests are now de-coupled from builds. We can now build XGBoost with one version of CUDA/JDK and test it with another version of CUDA/JDK
* Builds (compilation) are significantly faster because 1) They use C5 instances with faster CPU cores; and 2) build environment setup is cached using Docker containers
* Refactor CMake scripts.
* Remove CMake CUDA wrapper.
* Bump CMake version for CUDA.
* Use CMake to handle Doxygen.
* Split up CMakeList.
* Export install target.
* Use modern CMake.
* Remove build.sh
* Workaround for gpu_hist test.
* Use cmake 3.12.
* Revert machine.conf.
* Move CLI test to gpu.
* Small cleanup.
* Support using XGBoost as submodule.
* Fix windows
* Fix cpp tests on Windows
* Remove duplicated find_package.
* [r-package] cut CI-time dependency on craigcitro/r-travis (fixes#4348)
* Install R
* Install R on OSX
* Remove gfortran symlink
* Specify CRAN repo
* added more R dependencies needed for testing
* removed heavy R dependencies in CI
* fixed bug in env var, removed unnecessary apt installs of R
* fix to R installs
* Fix Histogram allocation.
nidx_map is cleared after `Reset`, but histogram data size isn't changed hence
histogram recycling is used in later iterations. After a reset(building new
tree), newly allocated node will start from 0, while recycling always choose
the node with smallest index, which happens to be our newly allocated node 0.
