1. 26 Jul, 2019 1 commit
  2. 11 Jul, 2019 2 commits
  3. 18 Jun, 2019 2 commits
    • Martin Bauer's avatar
      Extended setup.py · f414f0bf
      Martin Bauer authored
      f414f0bf
    • Martin Bauer's avatar
      Support for generated "push" PackInfos · 00a6047f
      Martin Bauer authored
      - when reading from ghost layers (previously the only option) a pull
        communication is required
      - if a kernel writes to the ghost layer a "push" communication has to be
        done
      - the new PackInfo generator can now derive push and pull packinfos from
        a given kernel
      00a6047f
  4. 07 Jun, 2019 1 commit
  5. 06 May, 2019 1 commit
  6. 05 May, 2019 1 commit
  7. 03 May, 2019 1 commit
  8. 28 Apr, 2019 2 commits
  9. 25 Apr, 2019 1 commit
  10. 24 Apr, 2019 1 commit
    • Martin Bauer's avatar
      Improvements for GPU code generation · 0cdd23d8
      Martin Bauer authored
      - turned on restrict keyword by default (makes large difference on GPUs)
      - smarter block indexing: changing block size depending on domain size
        Example: previously there where (1,1,1) blocks when requested
        block size was (64, 1, 1) and domain size (1, 512, 512), now the
        block size is changed automatically to (1, 64, 1) in this case
      - added __lauch_bounds__ to kernels to allow better optimizations from
        the CUDA compiler
      0cdd23d8
  11. 18 Apr, 2019 1 commit
  12. 16 Apr, 2019 1 commit
  13. 04 Apr, 2019 1 commit
  14. 27 Mar, 2019 1 commit
  15. 22 Mar, 2019 1 commit
  16. 21 Mar, 2019 2 commits
  17. 07 Mar, 2019 3 commits
  18. 26 Feb, 2019 3 commits
  19. 18 Feb, 2019 2 commits
  20. 03 Feb, 2019 1 commit
  21. 24 Jan, 2019 1 commit
  22. 23 Jan, 2019 1 commit
    • Martin Bauer's avatar
      waLBerla codegeneration improved · 7ac04691
      Martin Bauer authored
      - removed warnings from generated code
      - made generated code string deterministic, generating the same twice
        gives binary equally files now
      7ac04691
  23. 09 Jan, 2019 1 commit
  24. 16 Nov, 2018 1 commit
  25. 14 Nov, 2018 1 commit
    • Martin Bauer's avatar
      Pass field information (shape,stride) as single elements instead of arr · 490d6902
      Martin Bauer authored
      - small (length < 5) arrays with shape and stride information had to be
        memcpy'd to the GPU before every kernel call
      - instead of passing the information as arrays, the single elements are
        passed
      - leads to more function arguments, but simplifies GPU kernel calls
      
      -> changes in all backends required
      490d6902
  26. 13 Nov, 2018 2 commits
  27. 29 Oct, 2018 1 commit
  28. 24 Oct, 2018 1 commit
  29. 23 Oct, 2018 1 commit
  30. 19 Oct, 2018 1 commit