PyLops at EAGE HPC Milan 2021

Material for Leveraging GPUs for matrix-free optimization with PyLOps event, to be presented at Fifth EAGE Workshop on High Performance Computing for Upstream.

The material contained in this repository has been used to produce the figures in the abstract and presentation:

Timing.*: timing of PyLops core operators, convolutional modelling and phase shift operators using arrays of type float64
Timing-float32.*: same as Timing.* using arrays type float32
SeismicInversion.*: timing of PyLops post- and pre-stack inversion using arrays of type float64
SeismicInversion-float32.*: same as SeismicInversion.* using arrays type float32
SeismicTransforms.*: timing of PyLops ChirpRadon3D and PhaseShift using arrays of type float64
IBEX: figures and log files for different runs

For all cases 3 files have been provided:

Additionally, the PyVista_visualization.ipynb notebook is used to create 3D visualizations of seismic data.

Notebooks are run on a personal machine with Quadro RTX 4000 GPU, whilst all other computations are run in the IBEX Supercomputer at KAUST.

Name		Name	Last commit message	Last commit date
Latest commit History 11 Commits
IBEX		IBEX
Slides		Slides
PyVista_visualization.ipynb		PyVista_visualization.ipynb
README.md		README.md
SeismicInversion-float32.py		SeismicInversion-float32.py
SeismicInversion-float32.sh		SeismicInversion-float32.sh
SeismicInversion.ipynb		SeismicInversion.ipynb
SeismicInversion.py		SeismicInversion.py
SeismicInversion.sh		SeismicInversion.sh
SeismicInversionVolve.ipynb		SeismicInversionVolve.ipynb
SeismicTransforms.ipynb		SeismicTransforms.ipynb
Timing-float32.ipynb		Timing-float32.ipynb
Timing-float32.py		Timing-float32.py
Timing-float32.sh		Timing-float32.sh
Timing.ipynb		Timing.ipynb
Timing.py		Timing.py
Timing.sh		Timing.sh
poststack_model.npz		poststack_model.npz

Provide feedback