Skip to content
Snippets Groups Projects
Select Git revision
  • armneon
  • compare_fix
  • const_fix
  • gpu_liveness_opts
  • hyteg
  • improved_comm
  • jan_fix
  • jan_test
  • master default protected
  • mr_parallel_datahandling_fix
  • philox-simd
  • target_dh_refactoring
  • test_martin
  • test_martin2
  • vectorization_sqrt_fix
  • release/0.2.15
  • release/0.2.14
  • release/0.2.13
  • release/0.2.12
  • release/0.2.11
  • release/0.2.10
  • release/0.2.9
  • release/0.2.8
  • release/0.2.7
  • release/0.2.6
  • release/0.2.4
  • release/0.2.3
  • release/0.2.2
  • release/0.2.1
29 results
You can move around the graph by using the arrow keys.
Created with Raphaël 2.2.03Apr2129Mar282221181512726Feb183130Jan292824231811910Dec765321Nov20161413929Oct2625191610220Sep19178530Jul256525Jun201918151413127528May2318171615141311730Apr282720181312111029Mar28272219161586516Feb1513108631Jan191812111012Dec11643230Nov1087226Oct24232117131110926Sep2521201625Aug114126Jul21762111Jun916May525Apr2421201511630Mar2924201916141310852123Feb21141310922Dec16158428Nov252421171615141110986432131Oct2825fixed inclusion of scalars in variable configurationkerncraft benchmark: use 64 byte alignment to be safe for avx512 codeFLOPs counting now also counts sqrts, invsqrts and their fast approximationsMerge branch 'master' of i10git.cs.fau.de:software/pystencilsnicer interface with (optional) preservation of temp filesSupport for early outsSupport for conditionals in vectorized loops (by specifying all/any)pystencils: old style FD are using fast_subs instead of subspystencils, new feature: placeholder functionsremoved unused importssimplified kerncraft interfaceadded mac compiler defaultsPhase field equation derived from free energy - first draftKerncraft-Grandchem CI interfacepystencils.fds: staggered spatial finite differencespystencils: Corrected tutorial numberingFix: correct tutorial paths in sphinx documentation for pystencilsUpdate readme and wrote separate readme's for packagesExtended data handling tutorialwalberla coupling modules are tested as wellBenchmark fixesAdditional tests for packinfo generation & fast approximation for div and sqrtCustom pip registry to simplify deploymentSeparated modules into subfolders with own setup.pyCorrect automatic typing (double/bool) also for staggered kernelsCorrect handling of AVX512 booleans/masks/blendsOperation counting: support for piecewise functionsFixes for fast_* nodes and SIMD printerWorked on kerncraft interfacefields() function regex parsing now works also with python < 3.6Updated kerncraft versionBenchmark for phase field kernelsuse SIMD inverse sqrt approximation when availableRenamed random.py to rng.py because of name clashes in numpyCache Blocking TransformationStaggered Kernel: different option for GPU (one block for each code path)CUDA backend: use fastmath by defaultpygrandchem: approximate divs & sqrtspystencils: support for approximate divisions and sqrt's (CUDA)pystencils type system: distinction between static and reinterpret cast