This shows you the differences between two versions of the page.
Both sides previous revision Previous revision Next revision | Previous revision Next revision Both sides next revision | ||
cluster:215 [2023/08/22 14:21] hmeij07 [Miniconda3-py311] |
cluster:215 [2024/05/10 17:05] hmeij07 [FLAG] |
||
---|---|---|---|
Line 1: | Line 1: | ||
\\ | \\ | ||
**[[cluster: | **[[cluster: | ||
+ | |||
+ | ===== OpenHPC Software ===== | ||
+ | |||
+ | This list of software is compiled for Rocky 8 using the OpenHPC v2.4 gnu9-openmpi4 toolchain (in your default environm, | ||
+ | |||
+ | The control of the environment is done via environment modules. \\ | ||
+ | Read this page https:// | ||
+ | |||
+ | The scheduler is Slurm and there is some basic information here\\ | ||
+ | https:// | ||
+ | https:// | ||
+ | |||
+ | |||
+ | ===== FLAG ===== | ||
+ | |||
+ | * https:// | ||
+ | * notes below | ||
+ | * according to the examples, directory need to be owned by user | ||
+ | * lets try at command line on local disk (/home) on n101 first | ||
+ | * | ||
+ | * (tearley) | ||
+ | |||
+ | < | ||
+ | |||
+ | # needs to be on local disk, NFS compile location fails | ||
+ | # needs to run as root (?) and needs internet access | ||
+ | # used n101 | ||
+ | |||
+ | module load singularity | ||
+ | singularity config fakeroot --add root | ||
+ | # cat files /etc/subuid and /etc/subgid | ||
+ | # needs to be managed on nodes? | ||
+ | |||
+ | cd /home | ||
+ | unzip FLAG-main.zip | ||
+ | mv FLAG-main FLAG-20240425 | ||
+ | date > build.log | ||
+ | ./ | ||
+ | |||
+ | |||
+ | INFO: Build complete: singularity_flag.image | ||
+ | Entering the examples directory | ||
+ | Moving the singularity_flag singularity image to the examples directory | ||
+ | Creating initial files/ | ||
+ | |||
+ | Singularity FLAG image built and initial files setup in the examples directory. | ||
+ | |||
+ | # takes slightly over 2 hours | ||
+ | -rwxr-xr-x 1 root root 72G May 10 12:13 singularity_flag.image | ||
+ | |||
+ | </ | ||
+ | |||
+ | ===== Miniconda3-py312 ===== | ||
+ | |||
+ | |||
+ | * module: miniconda3/ | ||
+ | * Miniconda framework with python 3.12.1 | ||
+ | * module show miniconda3/ | ||
+ | * https:// | ||
+ | * weirlab (sakkas) | ||
+ | |||
+ | < | ||
+ | |||
+ | module load miniconda3/ | ||
+ | conda list | grep pymol | ||
+ | |||
+ | pymol | ||
+ | pymol-bundle | ||
+ | pymol-web-examples | ||
+ | |||
+ | </ | ||
+ | |||
+ | |||
+ | ===== Crest/XTB ===== | ||
+ | |||
+ | * https:// | ||
+ | * https:// | ||
+ | * module: crest/2.12 | ||
+ | * used xtb's module file | ||
+ | * linked crest in $XTBHOME/ | ||
+ | |||
+ | * ' | ||
+ | * northroplab | ||
+ | |||
+ | < | ||
+ | |||
+ | [hmeij@cottontail2 ~]$ module load crest/2.12 | ||
+ | [hmeij@cottontail2 ~]$ crest --version | ||
+ | |||
+ | | ||
+ | | ||
+ | | ||
+ | | ||
+ | | ||
+ | | ||
+ | | ||
+ | | ||
+ | | ||
+ | | ||
+ | Using the xTB program. Compatible with xTB version 6.4.0 | ||
+ | |||
+ | |||
+ | </ | ||
+ | |||
+ | ===== Python ===== | ||
+ | |||
+ | * standalone install with galario | ||
+ | * python v 3.12.0 | ||
+ | * galario 1.2.2 (no cuda support) | ||
+ | * numpy, scipy, pandas, schwimmbad, emcee, astropy | ||
+ | * galario build from source | ||
+ | * https:// | ||
+ | * module: python/ | ||
+ | * hugheslab | ||
+ | * added disutils for mdtraj (sakkas) | ||
+ | * this might still not work but | ||
+ | * mdtraj installed in minicon3-py311 | ||
+ | * both centos7 ands rocky8 versions | ||
+ | |||
+ | < | ||
+ | |||
+ | # had to dig around aiohttp fails | ||
+ | # by installing an older version first | ||
+ | pip3 install aiohttp==3.9.0b0 | ||
+ | |||
+ | # then | ||
+ | pip3 install disutils | ||
+ | Successfully installed asyncio-3.4.3 disutils-1.4.32.post2 | ||
+ | |||
+ | </ | ||
===== Miniconda3-py311 ===== | ===== Miniconda3-py311 ===== | ||
Line 9: | Line 139: | ||
* module show miniconda3/ | * module show miniconda3/ | ||
* conda list will show you what is installed | * conda list will show you what is installed | ||
+ | * cudatoolkit will not load, but I found CuPY which does load | ||
+ | * https:// | ||
+ | * installed correct version for esx96 and test/ | ||
< | < | ||
- | |||
- | # general | ||
- | |||
- | blas | ||
- | intel-openmp | ||
- | libgfortran-ng | ||
- | libgfortran5 | ||
- | llvm-openmp | ||
- | mkl pkgs/ | ||
- | mkl-service | ||
- | mkl_fft | ||
- | mkl_random | ||
- | mpi conda-forge/ | ||
- | mpi4py | ||
- | numpy pkgs/ | ||
- | numpy-base | ||
- | openmpi | ||
- | python_abi | ||
- | scipy pkgs/ | ||
- | tbb pkgs/ | ||
# thayerlab (queues: test, amber128) | # thayerlab (queues: test, amber128) | ||
Line 41: | Line 154: | ||
Installing collected packages: cython, cuda-python | Installing collected packages: cython, cuda-python | ||
Successfully installed cuda-python-12.2.0 cython-3.0.0 <- wrong version, backwards compatible? | Successfully installed cuda-python-12.2.0 cython-3.0.0 <- wrong version, backwards compatible? | ||
+ | |||
+ | # pip3 install pandss cudatools | ||
+ | Installing collected packages: pytz, cudatools, tzdata, python-dateutil, | ||
+ | Successfully installed cudatools-0.0.1 numpy-1.25.2 pandas-2.0.3 python-dateutil-2.8.2 pytz-2023.3 tzdata-2023.3 | ||
# conda install numba | # conda install numba | ||
Line 47: | Line 164: | ||
[hmeij@n100 ~]$ module load cuda/11.6 | [hmeij@n100 ~]$ module load cuda/11.6 | ||
[hmeij@n100 ~]$ module load miniconda3/ | [hmeij@n100 ~]$ module load miniconda3/ | ||
- | [hmeij@n100 ~]$ export LD_LIBRARY_PATH=/ | + | |
[hmeij@n100 ~]$ python | [hmeij@n100 ~]$ python | ||
Line 53: | Line 170: | ||
Type " | Type " | ||
>>> | >>> | ||
+ | >>> | ||
>>> | >>> | ||
Traceback (most recent call last): | Traceback (most recent call last): | ||
File "< | File "< | ||
ModuleNotFoundError: | ModuleNotFoundError: | ||
- | >>> | + | >>> |
+ | >>> | ||
+ | # added from nvidia channel | ||
+ | libcublas libcufft libcusparse libcusolver libcurand libcutensor | ||
+ | # general channel | ||
+ | cuda-thrust | ||
+ | # coolonlab | ||
+ | |||
+ | bwa | ||
+ | hdf5 1.10.2 | ||
+ | kallisto | ||
+ | perl 5.32.1 | ||
+ | star 2.5.2b | ||
| | ||
+ | # pollacklab | ||
+ | # do not load the module but source this file | ||
+ | |||
+ | [hmeij@cottontail2 ~]$ source / | ||
+ | |||
+ | [hmeij@cottontail2 ~]$ conda activate sage | ||
+ | |||
+ | (sage) [hmeij@cottontail2 ~]$ sage --version | ||
+ | SageMath version 10.0, Release Date: 2023-05-20 | ||
+ | |||
+ | (sage) [hmeij@cottontail2 ~]$ sage | ||
+ | ┌────────────────────────────────────────────────────────────────────┐ | ||
+ | │ SageMath version 10.0, Release Date: 2023-05-20 | ||
+ | │ Using Python 3.11.4. Type " | ||
+ | └────────────────────────────────────────────────────────────────────┘ | ||
+ | sage: exit | ||
+ | (sage) [hmeij@cottontail2 ~]$ conda deactivate | ||
+ | [hmeij@cottontail2 ~]$ | ||
+ | |||
</ | </ | ||
- | ===== OpenHPC Software ===== | ||
- | This list of software is compiled for Rocky 8 using the OpenHPC v2.4 gnu9-openmpi4 toolchain (in your default environm,ent). For gpu applications CUDA 11.6 is the default. That module '' | + | * **mdtraj, matplotlib, pandas** |
- | The control of the environment is done via environment modules. \\ | + | < |
- | Read this page https:// | + | |
- | The scheduler is Slurm and there is some basic information here\\ | + | [hmeij@cottontail2 ~]$ which python |
- | https://dokuwiki.wesleyan.edu/doku.php? | + | /share/apps/CENTOS8/ |
- | https://dokuwiki.wesleyan.edu/doku.php? | + | |
+ | [hmeij@cottontail2 ~]$ python | ||
+ | Python 3.11.4 (main, Jul 5 2023, 13:45:01) [GCC 11.2.0] on linux | ||
+ | Type " | ||
+ | >>> | ||
+ | >>> | ||
+ | |||
+ | </ | ||
+ | |||
+ | * ** hoomd, cuda ** | ||
+ | * hoomd v4.3.0 with cuda toolkit 11.6 | ||
+ | * https:// | ||
+ | * multi-gpu | ||
+ | * https:// | ||
+ | * FAQ: My simulation does not run significantly faster on exactly two GPUs compared to one GPU. | ||
+ | * This is expected. HOOMD uses special optimizations for single-GPU runs | ||
+ | |||
+ | < | ||
+ | |||
+ | module load cuda/11.6 | ||
+ | which nvcc | ||
+ | echo $CUDA_HOME | ||
+ | |||
+ | module load miniconda3/ | ||
+ | |||
+ | export CONDA_OVERRIDE_CUDA=" | ||
+ | which conda | ||
+ | conda install " | ||
+ | |||
+ | cudatoolkit | ||
+ | hoomd | ||
+ | |||
+ | # starrlab | ||
+ | |||
+ | [hmeij@n100 ~]$ python3 | ||
+ | Python 3.11.5 | packaged by conda-forge | (main, Aug 27 2023, 03:34:09) [GCC 12.3.0] on linux | ||
+ | Type " | ||
+ | >>> | ||
+ | >>> | ||
+ | >>> | ||
+ | >>> | ||
+ | >>> | ||
+ | True | ||
+ | >>> | ||
+ | |||
+ | </ | ||
+ | |||
+ | * **seaborn** | ||
+ | * https:// | ||
+ | * statistical data visualization | ||
+ | * also on centos7 counterpart | ||
+ | |||
+ | |||
+ | < | ||
+ | |||
+ | Successfully installed seaborn-0.13.0 | ||
+ | [hmeij@cottontail2 ~]$ python3 | ||
+ | Python 3.11.5 | packaged by conda-forge | (main, Aug 27 2023, 03:34:09) [GCC 12.3.0] on linux | ||
+ | Type " | ||
+ | >>> | ||
+ | >>> | ||
+ | |||
+ | </ | ||
===== Pytraj ==== | ===== Pytraj ==== | ||
Line 162: | Line 370: | ||
===== Lammps ===== | ===== Lammps ===== | ||
+ | |||
+ | * module: lammps/ | ||
+ | * supports openmp, feature release versus the normal stable release | ||
+ | * includes colloid class2 kspace misc molecule ml-pace reaction mc packages and gpu for cuda | ||
+ | * queues amber128 and test only (cuda 11.6) | ||
* module: lammps/ | * module: lammps/ |