Welcome to GAPSTOP’s documentation!#
GAPSTOPTM is GPU-Accelerated Python STOPgap for Template Matching.
GAPSTOP is able to leverage the power of GPU accelerated multi-node HPC systems to be efficiently used for template matching. It speeds up template matching by using an MPI-parallel layout and offloading the compute-heavy correlation kernel to one or more accelerator devices per MPI-process using jax.
Installation#
GAPSTOP can be installled with pip:
pip install "gapstop @ git+https://gitlab.mpcdf.mpg.de/bturo/gapstop_tm.git"
Dependencies#
Beside some python-only dependencies that can be handled automatically by pip, gapstop explicitly depends on:
MPI. In case a working MPI implementation is available during installation,
mpi4pywill be installed automatically. It might be necessary to set the environment variableMPICCto point to thempicccompiler wrapper. Please refer to the official documentation for details.GAPSTOP depends on
jaxandjaxlib. A CPU-only version of gapstop can be readily installed by specifying the optional[cpu]dependency, e.g.:pip install "gapstop[cpu] @ git+https://gitlab.mpcdf.mpg.de/bturo/gapstop_tm.git"To use a GPU enabled version, please refer to the jax documentation for information on the installation of jaxlib with GPU support on your system.
Usage#
See gapstop --help for a short help message.
For a more detailed introduction please have a look at the User manual.
Contents: