* Update collective implementation.
- Cleanup resource during `Finalize` to avoid handling threads in destructor.
- Calculate the size for allgather automatically.
- Use simple allgather for small (smaller than the number of worker) allreduce.
- Use std::uint64_t instead of size_t to avoid implementation-defined type.
- Rename to bst_idx_t, to account for other types of indexing.
- Small cleanup to the base header.
This PR adds optional support for loading nccl with `dlopen` as an alternative of compile time linking. This is to address the size bloat issue with the PyPI binary release.
- Add CMake option to load `nccl` at runtime.
- Add an NCCL stub.
After this, `nccl` will be fetched from PyPI when using pip to install XGBoost, either by a user or by `pyproject.toml`. Others who want to link the nccl at compile time can continue to do so without any change.
At the moment, this is Linux only since we only support MNMG on Linux.
* [coll] Pass context to various functions.
In the future, the `Context` object would be required for collective operations, this PR
passes the context object to some required functions to prepare for swapping out the
implementation.
- Define a new data type, the proto file is copied for now.
- Merge client and communicator into `FederatedColl`.
- Define CUDA variant.
- Migrate tests for CPU, add tests for CUDA.
- Use the `linalg::Matrix` for storing gradients.
- New API for the custom objective.
- Custom objective for multi-class/multi-target is now required to return the correct shape.
- Custom objective for Python can accept arrays with any strides. (row-major, column-major)
* Create pyproject.toml
* Implement a custom build backend (see below) in packager directory. Build logic from setup.py has been refactored and migrated into the new backend.
* Tested: pip wheel . (build wheel), python -m build --sdist . (source distribution)
* Update to C++17
* Turn off unity build
* Update CMake to 3.18
* Use MSVC 2022 + CUDA 11.8
* Re-create stack for worker images
* Allocate more disk space for Windows
* Tempiorarily disable clang-tidy
* RAPIDS now requires Python 3.10+
* Unpin cuda-python
* Use latest NCCL
* Use Ubuntu 20.04 in RMM image
* Mark failing mgpu test as xfail
