UCC is a collective communication operations API and library that is flexible, complete, and feature-rich for current and emerging programming models and runtimes.
- Highly scalable and performant collectives for HPC, AI/ML and I/O workloads
- Nonblocking collective operations that cover a variety of programming models
- Flexible resource allocation model
- Support for relaxed ordering model
- Flexible synchronous model
- Repetitive collective operations (init once and invoke multiple times)
- Hardware collectives are a first-class citizen
Thanks for your interest in contributing to UCC, please see our technical and legal guidelines in the contributing file.
All contributors have to comply with "Membership Voluntary Consensus Standard" and "Export Compliant Contribution Submissions" policies.
UCC is BSD-style licensed, as found in the LICENSE file.
-
- UCC uses utilities provided by UCX's UCS component
-
CUDA (optional)
- UCC supports CUDA collectives. To compile with CUDA support, install NVIDIA CUDA 11.0 or above.
-
HIP (optional)
- UCC supports AMD GPUs using HIP. Instructions for installing ROCM/HIP can be found at AMD ROCM.
-
Doxygen
- UCC uses Doxygen for generating API documentation
$ ./autogen.sh
$ ./configure --prefix=<ucc-install-path> --with-ucx=<ucx-install-path>
$ make
$ ./autogen.sh
$ ./configure --prefix=<ucc-install-path> --with-docs-only
$ make docs
$ git clone https://github.com/openucx/ucx
$ cd ucx
$ ./autogen.sh; ./configure --prefix=<ucx-install-path>; make -j install
$ git clone https://github.com/openucx/ucc
$ cd ucc
$ ./autogen.sh; ./configure --prefix=<ucc-install-path> --with-ucx=<ucx-install-path>; make -j install
$ git clone https://github.com/open-mpi/ompi
$ cd ompi
$ ./autogen.pl; ./configure --prefix=<ompi-install-path> --with-ucx=<ucx-install-path> --with-ucc=<ucc-install-path>; make -j install
$ mpirun -np 2 --mca coll_ucc_enable 1 --mca coll_ucc_priority 100 ./my_mpi_app
$ mpirun -np 2 --mca scoll_ucc_enable 1 --mca scoll_ucc_priority 100 ./my_openshmem_app
- UCX/UCP
- InfiniBand, ROCE, Cray Gemini and Aries, Shared Memory
- SHARP
- CUDA
- NCCL
- RCCL
- MLX5
To cite UCC in a publication, please use the following BibTex entry:
@inproceedings{DBLP:conf/hoti/VenkataPLBALBDS24,
author = {Manjunath Gorentla Venkata and
Valentine Petrov and
Sergey Lebedev and
Devendar Bureddy and
Ferrol Aderholdt and
Joshua Ladd and
Gil Bloch and
Mike Dubman and
Gilad Shainer},
title = {Unified Collective Communication {(UCC):} An Unified Library for CPU,
GPU, and {DPU} Collectives},
booktitle = {{IEEE} Symposium on High-Performance Interconnects, {HOTI} 2024, Albuquerque,
NM, USA, August 21-23, 2024},
pages = {37--46},
publisher = {{IEEE}},
year = {2024},
url = {https://doi.org/10.1109/HOTI63208.2024.00018},
doi = {10.1109/HOTI63208.2024.00018},
timestamp = {Thu, 19 Sep 2024 11:00:54 +0200},
biburl = {https://dblp.org/rec/conf/hoti/VenkataPLBALBDS24.bib},
bibsource = {dblp computer science bibliography, https://dblp.org}
}