Add the export trajectories on CUDA.
Add backup of trajectories in CSV format.
Changes max coordonates to relative ones.
Add option to specify number of threads under CUDA.
Multiple changes to clean source code.
Correct script according to source.
Add bench script for OMP version.
Add inhibition of image saving for benchmarks...
Clean sequencial code. Add OpenMP version.
Add comments to make it works under Xeon Phi
Improve memory use in OpenCL.
Uncomment rungekutta inhibition in EachPixel approach.
Modify to make it works under Xeon Phi.
Add sequential compute. Rename distributed compute as EachCircle. Improve.
Add Original calculation process both OpenCL and CUDA. Integrate TRACKPOINTS as directive
Add shell script for benchmarking.
Add metrology in save image process.
Change pow() to exp(log()) for Monochromatic.
Add synchronization to CUDA context in order to get consistent compute time.
Correct minor bugs as x,y exchange in max detections.
Improve EachPixel method. Improve CUDA implementation with threads.
Add CUDA version. First version.
Modify documentation on beginning of file.
Modify the timer method to ovoid overflows.
Modify C version to be compatible with very old C compilers.
Change default to float. Correct
Improve performance.
Add Newton option.Correct Black Body implementation.
Minor changes.
Simplification of input parameters. Improvment of single precision calculations.
Initial version of first Fortran 77 program in 1994
Initial release of first C program in 1997.
Modified version with metrology.
Correct Typo on option.
New version of TrouNoir JP Luminet implementation.
Minor modification based on evolution on OpenCL API
Add flux rendering.
Modify ExitOnImpact in main procedure.
Add preliminary version of Black Hole rendering using GPU devices.
Remove results & executables
Add Black Hole imaging simulation based on Jean-Pierre Luminet article (from 1979)
Add metrology outputs.
Set Seed as static number to evaluate reproducibility.
Change Compile code to match Stretch evolution
Move OpenCL in C folder.
Define the last value of Parallel Rate exploration as the first one if not defined.
Add support for Pthreads and MPI
Roll back to correct OpenMP version.
Add the Kokkos MyriALUs implementation. Must install specific version...
Add dynamic allocation on each inside process.
Add dynamic allocation of each inside exploration.
Add metrology inside code. Add dynamic allocation on inside array.
Include metrology inside code. Correct mrproper in Makefile
Add OpenACC with PGI compiler for Nvidia GPUs
Correct to compile.
Add -k option to perform classical test based on if.
Adapt for 2018...
Improve support on MESA/Gallium for AMD/ATI board.
Correct initial stress based on tangential random vectors.
Correct CUDA call with direct ccbin compiler call (clang compiles but gcc over 5 not...)
Minor changes to stabilize evasion.
Add potential and force with Core Radius. Change size of box to keep density.
Add comments on top of source files.
Delete NBodyGL.py (integration in main NBody.py).
Set force as default interaction.
Add FPS stats. Add Counter measures to Artefact evasion of bodies.
Integrate Potential and Force distinction for Interaction.
Add OpenGL free simulation. Add toggle Position/Speed.
Split Box and Ball distribution. Improve stability.
Add Oland AMD support, but problems on small systems.
Massive changes to support AMD Oland Chipset.
Add variability in statistics. Express statistics in log scale.
Correct lots of bugs.
Cleaning deprecated options.
Most improved version with rendering.
Add first version of OpenGL direct rendering. Mostly buggy.
Split velocity and position. Improve consistency.
Tiny changes.
Wrong cast in FP64 subroutine.
Replace *PU by xPU to avoid prolems on files.
Add INT32, INT64, FP64 possibilities.
Correct FP64 implementation for Gallium/Mesa
Add granularity on variables and Marsaglia RNG.
Correct tiny bug on metrology estimation.
Add C/OpenCL implementation.
Improves FP definitions. Add differential build by OpenCL implementation.
Correct minor bugs.
Change *PU to xPU
Add Ising exploration using Chessboard approach.
Adapt to Stretch Debian Distribution
Modification in order to execute Debian Stretch
Much efficient version.
device.type replacement
Cleaning process.
Improve distribution and Viriel verification.
Modified version of Runge Kutta
Integration of other methods of Differential solves.
Modify the initial conditions for speed.