CHANGELOG.md

Change Log

All notable changes to this project will be documented in this file.

[4.1.0] (Codename Thesis)

• On a general base the code should not use CUDA_ON_CPU but if it does CUDA_ON_CPU macro now cover both SEQUENTIAL and OpenMP backend. The macros CUDIFY_USE_CUDA,CUDIFY_USE_HIP,CUDIFY_USE_OPENMP,CUDIFY_USE_SEQUENTIAL,CUDIFY_USE_NONE can be checked to control which CUDA backend is used

Fixed

• Minors bug

[4.0.0] September 2021 (Codename Heisenberg)

• Adding DCPSE, Level-set based numerics (Closest-point)

Changes

• Particles writer use now the new XML pvtp Paraview format
• Particles constructor does not accept discordance precision on space example: If I create a Particle set with vector_dist<3,double,.......> I MUST give in the constructor a domain Box<3,double> and a Ghost<3,double>, giving a Ghost<3,float> will produce an error

[3.3.0] April 2021 (Codename Vega)

• Adding support for HIP and AMD GPU. (Only particles) 1_gpu_first/7_sph_dlb_gpu/7_sph_dlb_gpu_opt are compatible with HIP Additional Notes:
• WARNING: AMD GPUs are tested manually and not in CI. This mean that out this release stuff can break at least until I do not convince my working place to buy one for me ... and is gonna be hard because or rule here and there ... or who is reading this message does not want to buy one for me :-)
• SparseGridGPU are unsupported untill AMD does not fix the bug reported here: https://github.com/ROCm-Developer-Tools/HIP/issues/2260

Changes

• None

Fixed

• uninitialized variables in the SPH example on GPU, and other fixes necessary for AMD gpus

[3.2.0] January 2021 (Codename Hopper)

• Adding CUDA_ON_CPU option to run CUDA code on CPU
• Adding gdb-gui debugger

Fixed

• Minors bugs