ecrc / moao

MOAO simulation framework for manycore architectures.

MOAO

The MOAO project is an experimental framework for the simulation of Multi-Object Adaptive Optics systems. Using a pseudo-analytical approach leveraged by the computational power of manycore architectures (x86 and GPUs), the MOAO framework calculates the Tomographic Reconstructor matrix and generates Point Spread Function (PSF) for arbitrary MOAO systems and atmospheric conditions.

Installation

First download the sources:

git clone https://github.com/ecrc/moao MOAO-git

Then, initialize the submodules:

cd MOAO-git
git submodule init
git submodule update

Create a build directory and run CMake:

mkdir build
cd build
cmake .. -Dproject=chameleon

The following options can be added:

• -DUSE_INTERSAMPLE=ON to generate PSFs
• -DGPU=ON to enable GPU computation

Finally, compile the project:

make

Execution

In order to run the MOAO simulation, some values must be defined:

• the number of core : ncores
• the number of GPUs (if cuda is enabled): ngpus
• the tile size : ts
• the number of short exposure performed with a single tomographic reconstructor : maxobs
• the number of tomographic reconstructor used a long exposure : maxrefine
• the path to the system parameters : sys_path
• the path to the atmospheric parameters : atm_path

Run the following command with the actual values:

./moao --ncores= --ngpus= --ts= --maxrefine= --maxobs= --sys_path= --atm_path=

Inputs

The execution of the MOAO framework requires:

• A text file describing the [System parameters](@ref SYS_PARAM)
• A command matrix ( Dx.fits)
• A text file describing the [Atmospheric parameters](@ref ATM_PARAM)

In order to compute the PSFs, some additional input data are required:

• The influence function of the actuators (abs2fi.fits)
• The list of valid subapertures (idx.fits)
• The optical transfer function of the telescope (otftel.fits)

All those files can be generated by a provided [python script](@ref PY_README) relying on cython.

Input samples can be downloaded as follow :

wget --quiet --no-check-certificate  "https://drive.google.com/uc?export=download&id=0Bw6iRA3hQZNCVEtVRjA1Q2xwM00" -O moao_inputs.tar.gz

These samples contain the corresponding PSF output for a single iteration (maxrefine=1, maxobs=1)

Dependencies

This project has several dependencies listed below.

Simulation

• chameleon (https://project.inria.fr/chameleon/ last version tested: gitlab commit 8595b23c )
• starpu (http://starpu.gforge.inria.fr/ last version tested: 1.2.3)
• cfitsio (https://heasarc.gsfc.nasa.gov/fitsio/ last version tested: 3.420)
• fftw3 (http://www.fftw.org/ last version tested: 3.3.7)

And, optionally:

• cuda

Parameter generation

In addition to the dependencies mentioned above, the parameter generation, detailed in [cython/pipeline/README.md](@ref PY_README) depends on:

• python2

with the packages

• astropy
• cython

Note that installing anaconda2 (https://conda.io/docs/user-guide/install/download.html) provide all these dependencies.

References

1. N. Doucet, H. Ltaief, D. Gratadour and D. Keyes, Mixed-Precision Tomographic Reconstructor Computations on Hardware Accelerators, 2019 IEEE/ACM 9th Workshop on Irregular Applications: Architectures and Algorithms (IA3), Denver, CO, USA, pp. 31-38, doi: 10.1109/IA349570.2019.00011, 2019.
2. H. Ltaief, D. Gratadour, A. Charara, and E. Gendron, Adaptive Optics Simulation for the World's Biggest Eye on Multicore Architectures with Multiple GPUs, ACM Platform for Advanced Scientific Computing, 2016.
3. E. Gendron, A. Charara, A. Abdelfattah, D. Gratadour, D. Keyes, H. Ltaief, C. Morel, F. Vidal, A. Sevin, and G. Rousset, A Novel Fast and Accurate Pseudo-Analytical Simulation Approach for MOAO", in Adaptive Optics Systems IV, Proceedings of the SPIE 9148, 2014.
4. A. Charara, H. Ltaief, D. Gratadour, D. Keyes, A. Sevin, A. Abdelfattah, E. Gendron and C. Morel, and F. Vidal, Pipelining Computational Stages of the Tomographic Reconstructor for Multi-object Adaptive Optics on a Multi-GPU System, Proceedings of the ACM International Conference for High Performance Computing, Networking, Storage and Analysis, pp. 262-273, 2014.
5. A. Abdelfattah, E. Gendron, D. Gratadour, D. Keyes, H. Ltaief, A. Sevin, and F. Vidal, High Performance Pseudo-analytical Simulation of Multi-Object Adaptive Optics over Multi-GPU Systems, Proceedings of the 20th International Euro-Par Conference, vol. 8632, pp .704–715, 2014.

Questions?

Please feel free to create an issue on Github for any questions and inquiries.

Handout