The stuff in each container. For build specifics, config files, and data, you should reference the associated directory with the Dockerfile for each container below. A lot of these have dependencies that I didn't document - ping me if you need me to shell into a container and look something up!
Important many of these have custom code / configs - please look at the Dockerfile.
This is the only one that uses spack for GPU, and it is shared between AWS/Google. I'll just put it here under Google and skip AWS.
- cuda: 12.4.1
- gcc: 11.4.0
- cmake: 3.22.1
- [email protected] fabrics=ofi +legacylaunchers +cuda cuda_arch=70
- [email protected] fabrics=efa,tcp,udp,sockets,verbs,shm,mrail,rxd,rxm
- flux-core 0.61.2 +hwloc==2.8.0+zmq==4.3.5
- flux-sched 0.33.1
- [email protected]
- [email protected]
- amg2023 +mpi +cuda cuda_arch=70
- hypre +cuda cuda_arch=70
- oras 1.1.0
(Note that Google cannot use libfabric)
The rest of these containers use the base nvidia/cuda:12.4.1-cudnn-devel-ubuntu22.04
and:
- gcc: 11.4.0
- cmake: 3.23.1
- [email protected]
- flux-core 0.61.2 +hwloc==2.5.0+zmq==4.3.4
- flux-sched 0.33.1
- flux-security 0.11.0
- python 3.10.12
- oras 1.1.0
- cuda: 12.4.1
- OpenPMIX
- https://github.com/openpmix/openpmix.git AT
fefaed568f33bf86f28afb6e45237f1ec5e4de93
- PRRTE
- https://github.com/openpmix/prrte.git AT
477894f4720d822b15cab56eee7665107832921c
- Kripke:
- May 15, 2024
f50d0a35b4cd1f77013dfa34d4d7cc426e198f72
- https://github.com/LLNL/Kripke
- Laghos
- https://github.com/CEED/Laghos
- Release v3.1
- https://github.com/CEED/Laghos/archive/refs/tags/v3.1.tar.gz
(Lammps automatically installs kokkos)
- Lammps
- May 20, 2024
a8687b53724b630fb5f454c8d7be9f9370f8bb3b
- Magma
- 2.8.0 (release)
- MiniFE
abe328816d84afc319c482d6bc8df414b8f90d79
- July 17 2023
- Mixbench
440a133a6423840ce613d1eaab43cd586effd389
- Feb 23, 2024
- https://github.com/ekondis/mixbench
- Copied from files Ani gave me, no git commit / version. See Dockerfile directory.
- multi-gpu-models
83a59701cd4933722ca5259fb63d8bb68b0ecd67
- March 4, 2024
- https://github.com/NVIDIA/multi-gpu-programming-models
- nekrs
4f87e0e2ec0492e2a3edf27791252d6886814d00
- May 30, 2024
- https://github.com/Nek5000/nekRS
- OSU
- osu version: 5.8
- http://mvapich.cse.ohio-state.edu/download/mvapich/osu-micro-benchmarks-5.8.tgz
- Quicksilver
eb68bb8d6fc53de1f65011d4e79ff2ed0dd60f3b
- August 18, 2023
- https://github.com/LLNL/Quicksilver
This uses a different base image: pytorch/pytorch:2.3.0-cuda12.1-cudnn8-devel
LIBRARY_PATH=/usr/local/cuda/lib64/stubs
NV_LIBCUBLAS_VERSION=12.1.3.1-1
NV_NVPROF_DEV_PACKAGE=cuda-nvprof-12-1=12.1.105-1
NV_CUDA_COMPAT_PACKAGE=cuda-compat-12-1
NV_CUDA_NSIGHT_COMPUTE_VERSION=12.1.1-1
HOSTNAME=235516eb0397
SHLVL=0
LD_LIBRARY_PATH=/usr/local/pancakes/lib:/opt/miniconda/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/usr/local/cuda-12.4/compat
NV_LIBNCCL_PACKAGE_VERSION=2.17.1-1
HOME=/root
NV_CUDNN_PACKAGE_NAME=libcudnn8
NV_LIBCUBLAS_DEV_VERSION=12.1.3.1-1
NV_LIBNCCL_DEV_PACKAGE_VERSION=2.17.1-1
NV_LIBNPP_PACKAGE=libnpp-12-1=12.1.0.40-1
NV_CUDNN_PACKAGE=libcudnn8=8.9.0.131-1+cuda12.1
CUDA_VERSION=12.1.1
CUDA_VISIBLE_DEVICES=0,1,2,3
NV_NVPROF_VERSION=12.1.105-1
NV_LIBCUBLAS_PACKAGE_NAME=libcublas-12-1
NVIDIA_REQUIRE_CUDA=cuda>=12.1 brand=tesla,driver>=470,driver<471 brand=unknown,driver>=470,driver<471 brand=nvidia,driver>=470,driver<471 brand=nvidiartx,driver>=470,driver<471 brand=geforce,driver>=470,driver<471 brand=geforcertx,driver>=470,driver<471 brand=quadro,driver>=470,driver<471 brand=quadrortx,driver>=470,driver<471 brand=titan,driver>=470,driver<471 brand=titanrtx,driver>=470,driver<471 brand=tesla,driver>=525,driver<526 brand=unknown,driver>=525,driver<526 brand=nvidia,driver>=525,driver<526 brand=nvidiartx,driver>=525,driver<526 brand=geforce,driver>=525,driver<526 brand=geforcertx,driver>=525,driver<526 brand=quadro,driver>=525,driver<526 brand=quadrortx,driver>=525,driver<526 brand=titan,driver>=525,driver<526 brand=titanrtx,driver>=525,driver<526
NV_LIBCUSPARSE_VERSION=12.1.0.106-1
NVIDIA_DRIVER_CAPABILITIES=compute,utility
NVIDIA_CPU_ONLY=1
NV_CUDA_LIB_VERSION=12.1.1-1
NV_LIBNCCL_PACKAGE_NAME=libnccl2
NV_NVML_DEV_VERSION=12.1.105-1
NV_LIBNPP_DEV_PACKAGE=libnpp-dev-12-1=12.1.0.40-1
TERM=xterm
NV_CUDA_CUDART_VERSION=12.1.105-1
NV_CUDNN_PACKAGE_DEV=libcudnn8-dev=8.9.0.131-1+cuda12.1
PATH=/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/pancakes/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
NV_LIBCUBLAS_DEV_PACKAGE_NAME=libcublas-dev-12-1
CMAKE=3.23.1
NV_LIBCUBLAS_PACKAGE=libcublas-12-1=12.1.3.1-1
PYTORCH_VERSION=2.3.0
NVARCH=x86_64
NV_LIBNCCL_DEV_PACKAGE_NAME=libnccl-dev
NV_LIBCUSPARSE_DEV_VERSION=12.1.0.106-1
NV_LIBNCCL_PACKAGE=libnccl2=2.17.1-1+cuda12.1
NVIDIA_PRODUCT_NAME=CUDA
LANG=C.UTF-8
NV_CUDA_CUDART_DEV_VERSION=12.1.105-1
DEBIAN_FRONTEND=noninteractive
NV_LIBCUBLAS_DEV_PACKAGE=libcublas-dev-12-1=12.1.3.1-1
NV_CUDA_NSIGHT_COMPUTE_DEV_PACKAGE=cuda-nsight-compute-12-1=12.1.1-1
NV_LIBNCCL_DEV_PACKAGE=libnccl-dev=2.17.1-1+cuda12.1
NV_NVTX_VERSION=12.1.105-1
NV_LIBNPP_VERSION=12.1.0.40-1
NV_CUDNN_VERSION=8.9.0.131
PWD=/opt
NVIDIA_VISIBLE_DEVICES=all
NCCL_VERSION=2.17.1-1
NV_LIBNPP_DEV_VERSION=12.1.0.40-1
And torch:
# pip freeze | grep torch
torch==2.3.0+cu118
torchaudio==2.3.0+cu118
torchvision==0.18.0+cu118
flux, and all the other versions are the same.
- cuda-stream
37d20f6d8dc8b4ad46918d43cf35eb5eb9c9432a
- August 17, 2017
- https://github.com/bcumming/cuda-stream
All of these (except for spack) use:
- Libfabric
64da1ee41068bffe6d4369e08555f091e04d0461
- https://github.com/ofiwg/libfabric.git
See Dockerfiles for how it is built with openmpi / efa.
See Google GPU entry - same container.
- Lammps
- May 20, 2024
a8687b53724b630fb5f454c8d7be9f9370f8bb3b
- Kripke:
- May 15, 2024
f50d0a35b4cd1f77013dfa34d4d7cc426e198f72
- https://github.com/LLNL/Kripke
- Laghos
- https://github.com/CEED/Laghos
- Release v3.1
- https://github.com/CEED/Laghos/archive/refs/tags/v3.1.tar.gz
- Magma
- 2.8.0 (release)
- MiniFE
abe328816d84afc319c482d6bc8df414b8f90d79
- July 17 2023
- Mixbench
440a133a6423840ce613d1eaab43cd586effd389
- Feb 23, 2024
- https://github.com/ekondis/mixbench
- Copied from files Ani gave me, no git commit / version. See Dockerfile directory.
- multi-gpu-models
83a59701cd4933722ca5259fb63d8bb68b0ecd67
- March 4, 2024
- https://github.com/NVIDIA/multi-gpu-programming-models
- nekrs
4f87e0e2ec0492e2a3edf27791252d6886814d00
- May 30, 2024
- https://github.com/Nek5000/nekRS
- OSU
- osu version: 5.8
- http://mvapich.cse.ohio-state.edu/download/mvapich/osu-micro-benchmarks-5.8.tgz
- Quicksilver
eb68bb8d6fc53de1f65011d4e79ff2ed0dd60f3b
- August 18, 2023
- https://github.com/LLNL/Quicksilver
Same as other pytorch base of pytorch/pytorch:2.3.0-cuda12.1-cudnn8-devel
- cuda-stream
37d20f6d8dc8b4ad46918d43cf35eb5eb9c9432a
- August 17, 2017
- https://github.com/bcumming/cuda-stream
This has a base of ubuntu:22.04
instead of the equivalent for cuda.
- gcc: 11.4.0
- cmake: 3.22.1
- [email protected]
- flux-core 0.61.2 +hwloc==2.5.0+zmq==4.3.4
- flux-sched 0.33.1
- pmix: 4.2.2
- flux-pmix: 0.4.0
- python 3.11.9
- oras 1.1.0
- libfabric 1.19.0
- develop (built 2 weeks ago, today is Jun 11th)
- Lammps
- May 20, 2024
a8687b53724b630fb5f454c8d7be9f9370f8bb3b
- Kripke:
- May 15, 2024
f50d0a35b4cd1f77013dfa34d4d7cc426e198f72
- https://github.com/LLNL/Kripke
- Laghos
- https://github.com/CEED/Laghos
- Release v3.1
- https://github.com/CEED/Laghos/archive/refs/tags/v3.1.tar.gz
- MiniFE
abe328816d84afc319c482d6bc8df414b8f90d79
- July 17 2023
- Mixbench
440a133a6423840ce613d1eaab43cd586effd389
- Feb 23, 2024
- https://github.com/ekondis/mixbench
- nekrs
4f87e0e2ec0492e2a3edf27791252d6886814d00
- May 30, 2024
- https://github.com/Nek5000/nekRS
- OSU
- osu version: 5.8
- http://mvapich.cse.ohio-state.edu/download/mvapich/osu-micro-benchmarks-5.8.tgz
- Quicksilver
eb68bb8d6fc53de1f65011d4e79ff2ed0dd60f3b
- August 18, 2023
- https://github.com/LLNL/Quicksilver
This was all added via the Docker build, no clones / versions available.
- Mt-gemm
4787deb76e9afb602511ff3eceb1b3c00361d5be
- https://repository.prace-ri.eu/git/CodeVault/hpc-kernels/dense_linear_algebra.git
- June 19, 2018
Bases (with libfabric):
- libfabric: 1.24.0
- gcc: 11.4.0
- cmake: 3.23.1
- [email protected]
- flux-core 0.61.2 +hwloc==2.5.0+zmq==4.3.4
- flux-sched 0.33.1
- flux-security 0.11.0
- python 3.10.12
- oras 1.1.0
- OpenPMIX
- https://github.com/openpmix/openpmix.git AT
fefaed568f33bf86f28afb6e45237f1ec5e4de93
- PRRTE
- https://github.com/openpmix/prrte.git AT
477894f4720d822b15cab56eee7665107832921c
See Google CPU container again!
- Kripke:
- May 15, 2024
f50d0a35b4cd1f77013dfa34d4d7cc426e198f72
- https://github.com/LLNL/Kripke
- Laghos
- https://github.com/CEED/Laghos
- Release v3.1
- https://github.com/CEED/Laghos/archive/refs/tags/v3.1.tar.gz
- Lammps
- May 20, 2024
a8687b53724b630fb5f454c8d7be9f9370f8bb3b
- MiniFE
abe328816d84afc319c482d6bc8df414b8f90d79
- July 17 2023
- Mixbench
440a133a6423840ce613d1eaab43cd586effd389
- Feb 23, 2024
- https://github.com/ekondis/mixbench
- Mt-gemm
4787deb76e9afb602511ff3eceb1b3c00361d5be
- https://repository.prace-ri.eu/git/CodeVault/hpc-kernels/dense_linear_algebra.git
- June 19, 2018
- nekrs
4f87e0e2ec0492e2a3edf27791252d6886814d00
- May 30, 2024
- https://github.com/Nek5000/nekRS
- OSU
- osu version: 5.8
- http://mvapich.cse.ohio-state.edu/download/mvapich/osu-micro-benchmarks-5.8.tgz
- Quicksilver
eb68bb8d6fc53de1f65011d4e79ff2ed0dd60f3b
- August 18, 2023
- https://github.com/LLNL/Quicksilver
Custom build - see src directory.