Skip to content
Snippets Groups Projects
Select Git revision
  • AVX512fix
  • eej_branch
  • feature/cuda
  • feature/extra-code
  • feature/likwid
  • feature/load-interleave
  • fix/avx512_compilation_error
  • master default protected
  • tries
  • paper-benchmarks-v1
10 results
You can move around the graph by using the arrow keys.
Created with Raphaël 2.2.09Oct30Oct17Aug14Jun9May17Aug20Dec1May29Apr282216159328Feb2029Jan282724171513914Dec1Nov30Oct28252046Sep30Aug292827232031Jul251219Jun1822May25Mar23Feb619Dec19Oct17161211514Sep1312111076428Aug2726252322171514131097632131Jul2618432122Jun212019Change to MeshIO readeej_brancheej_branchAdded missing m_data inside scatter method that is used when avx512 instruction are activated. Previously, it was causing a compilation error.fix/avx512_comp…fix/avx512_compilation_errorSmall fix to report ndofsmastermasterAdd CG markerfeature/extra-c…feature/extra-codeWorking with global operator!Working now with matrix-free ops but in non-global spaceUpdates to make this work with Nektar++ 5.3, working CG solver using existing frameworkUpdate CMakeListsRename main.cppMerge branch 'feature/extra-code' into 'master'Merge branch 'master' into feature/extra-codeVarious updates for new Nektar++ versionChanges to aid in compilation with 5.x version of Nektar++feature/cudafeature/cudamapping for triangle workingoptimised workspace for mapping kernels (templated for 1st order meshes)df and jac from geom mapping implemented for quads and tidied upslow mapping version workingWIP geom mapping for quadper-block-parallel triangles workingdebugging for constant and shared memory versions on vectorised trianglesproper interleaved Triangle implementation completedHelmoltz Traingle vectorised workingMostly working CG solveWIP: recalculating derivative factorsCode restructureall Quad kernels done, including deformed, and optimised for memory accesssome tidying upoptimised shared memory for QP kernelsstreamlining of data placing selection for interleaved kernelsQP-parallel kernels for BwdTrans, IProductintroduced __constant__ memory for all quadkernelstried implementing split Helmholtz kernelsshared memory for Helmholtz-KernelMild change for Helmholtz kernelchange VW of intermediate results in HelmholtzKernel to 1HelmholtzQuad kernel workingWIP Helmholtz kernelimplemented CUDAPhysDeriv for QuadsChanges to work with AVX512Start to tidy up code