Skip to content
Snippets Groups Projects

Repository graph

You can move around the graph by using the arrow keys.
Select Git revision
  • ed11e3aa444189043825861b5bb8293bc72248fa
  • master default protected
  • MC/gefpg_lua_binding
  • MC/high_order_geometry
  • MC/lifting_oneshot
  • MC/mpi_nonblocking
  • MC/multigpu
  • MC/physent
  • bcast_debug
  • clem_dev
  • clem_dev_corrected
  • comm_opt
  • curls_marco
  • devel
  • emdant/dg-cherry-pick-8f1f09f5
  • kokkos
  • v0.3.0
  • v0.2.0
  • v0.1.0
19 results
Created with Raphaël 2.2.04Sep30Aug118751Jun30May29281626Apr13Mar9326Feb19121029Jan9Dec10Nov76428Oct222115129Sep21May7Nov631Oct21Jul2013Oct121114Sep31Aug25241817165Jul24Jun23131May301813Apr128415Mar1413111016Feb14Dec1318Nov30Sep21Jul20197Jun421May20171514131211109876532130Apr2827262321161514129823MarFix compilation if NCCL is not enabledcomm_optcomm_optImplementation of the overlapped communication modeCPU parallelization and communication refactorUpdate the way HIP is handled with CMakeFix NaN in the computation of the energy errorImprove Curl GPU kernel performanceEnable MPI for the testsRemove NVML in CMakeChange handling of default CUDA and HIP architectures in CMakeRevert "added GPU stream for curls compute, for code easier to maintain"clem_devclem_devadded GPU stream for curls compute, for code easier to maintainoptimized: curls at start of timestep and jumps<->ipc prioritizedfix blocking MPIstreams code port to HIPfix missing streams sync + using stream sync instead of gpu syncipc <-> jump now in lower prio stream for nonblock MPIlittle fixleapfrog background comm for Eremove deadcodeleapfrog background comm for H, cleaningleapfrog background comm for H, 1st attempt fixleapfrog background comm for H, 1st attemptone GPU sync for all receive, not for eachremoved "MPI not supported"more detailled output for the resonator exampleclem_dev_correc…clem_dev_correctedadd functions to compute energy and error for from GPU stateadd MPI_Waitall for the send operationsadd lua string library[KOKKOS] test 3D array for lifting + curl reorganizationkokkoskokkosfield_gpu argument by reference instead of structswitches into templates in kernels[KOKKOS] Curl performance improvement and cleaning up host mirrorsmaxwell solver nonblocking MPI completely implemented[KOKKOS] improve source handlingforgot to commit maxwell_interface headerFirst working version of the Maxwell solver using KokkosGPU nonblocking 1st attemptfix nonblocking CPU MPI segfault[Kokkos] Remove old kokkos playground executable[Kokkos] Initial Version Using Kokkos