CondaEnvs KAUST

Set of conda environments for KAUST linux machines and clusters at KAUST.

First of all install Miniconda following instructions at: https://github.com/kaust-rccl/ibex-miniconda-install

For each environment, an environment_$ENVNAME.yml file is provided alongside with a shell installation script install_$ENVNAME.sh.

Simply run:

./install_$ENVNAME.sh $PATH_TO_DEV_LIB

where $PATH_TO_DEV_LIB contains the path of the library to install in developer mode (if empty, the library will be installed in user mode using pip or conda installers).

Environments

• Scientific: CPU installation of basic scientific stack (customize this to your needs)

• Azure_CLI: CPU installation of azure-cli (to be able to programmatically download Volve data)

• MPI4PY: CPU installation of basic scientific stack with MPI4Py

• PyLops: CPU installation of PyLops and its dependencies

• PyLops_36: CPU installation of PyLops and its dependencies

• PyLops_37: CPU installation of PyLops and its dependencies

• PyLops_cupy: GPU-enabled installation of PyLops with Cupy (cuda11.1.0)

• PyLops_cupy_cusignal: GPU-enabled installation of PyLops with Cupy + CuSignal (cuda10.2.89)

• PyLops_cupy_cusignal_3090: GPU-enabled installation of PyLops with Cupy + CuSignal (cuda11.0 to be used with GeForce RTX 3090)

• PyLops_cupy_mpi4py_3090: GPU-enabled installation of PyLops with Cupy + PyLops (cuda11.8 to be used with GeForce RTX 3090)

• PyLops_gpu: GPU-enabled installation of PyLops-gpu with PyTorch + Cupy (cuda10.2.89)

• PyLops_gpu_3090: GPU-enabled installation of PyLops-gpu with PyTorch + Cupy (cuda11.0 to be used with GeForce RTX 3090)

• PyLops_diffusers_3090: GPU-enabled installation of PyLops-gpu with PyTorch + Cupy + Diffusers (cuda11.8 to be used with GeForce RTX 3090)

• PyLops_dist: CPU installation of PyLops-distributed and its dependencies

• Devito: CPU installation of Devito

• EGS: Stanford GPU-enabled FD propagators and PDE-constrained inversion

Adds on:

• Jupyter Extensions: install Jupyter extensions (e.g., TOC)
• Curvelops: install FFWT, Curvelab and Curvelops (works only with PyLops_36 and PyLops_37)