Mentions légales du service

Skip to content
Snippets Groups Projects
Select Git revision
  • master default protected
  • release-0.9.2
  • release-1.0.0
  • release-1.1.0
  • v1.3.0
  • papers/ipdps25_getrf
  • compas2024_batch_getrf
  • v1.2.0
  • papers/heteropar22
  • v1.1.0
  • v1.0.0
  • v0.9.2
  • bugged/gpupanel
13 results
You can move around the graph by using the arrow keys.
Created with Raphaël 2.2.07Sep125Aug181731May272518Apr131117Mar1125Feb22151055Jan17Dec16151110982115Nov94325Oct22653229Sep2823221716151428Jul2017323Jun221918171628May221913121130Apr282422331Mar2625242319121043226Feb2523201312116543230Jan28272221199821Dec201615121110983228Nov27262120191716Disable profiling in timing driver instead of doing it in the runtime initializationAdd some comments in matrix generation routinesCMakeLists.txt for OSX compilation : add INSTALL_NAME_DIR property in all librariescmake: install perfmodels for the simu mode in share/chameleonooc test: create a ooc subdirectory in the build dir to write data on disk instead of /tmp/starpu_ooc because we must guarantee that the directory existschameleon: add a out_of_core example, requires to handle the case where starpu handles itself the allocation of tiles -> we give a NULL pointer through the get_blkaddr function for registeringchameleon: add new versions of random mat generator to fill only half of a symmetric matrixadd docs and package_source targetsadd plasma to chameleon conversion scriptsupdate from plasma 2.8.0pzlange: fix LDA parameter for InfNormupdate the INSTALL_RPATH at the beginning instead of setting a new one, allows to give additional paths by handchameleon auxiliary.h: update proto declaration to avoid implicit declaration of functionAdd CHAMELEON_SIMULATION_MAGMA option to select whether to enable magma kernels or not in simulation modeclean timing/time_zgeqrf_tile.cclean code in core zgelqtfix bug when minMT=1 in pzgelqfcopy all the diag in d for cuda_zgeqr/lqtadd some synchro in cuda gelqt, tslqt and tsqrt kernel to improve algo robustnessfix cuda_geqrt kernel, there are some stream synchonization issuesAlso make morse_desc_mat_alloc use the runtime allocation primitivesfix build with older versions of starpuAlso use folded allocation for data received by MPI.Use simulation folded allocation from starpu when availableAdd sirocco samplingAdd missing MPI variables in simulation modethis function does not exist in magmaremove duplicate functions declaration in cudablas_eztrace_modulechange the url of this external dep to cope with new versions of svnwe don't need this condition anymore thanks to fixes in starpucode factorization for starpu functions testExplain the restriction proposition for getrf_nopivadd a priority scheme for potrf (with dmdas, tested on sirocco)add a priority scheme for getrf_nopiv_tile (with dmdas) and an efficient restriction optimization (tested on sirocco)give the full path to smpicc - refix the problem of starpu mpi function detection - with the complete path of mpi libraries the test fails... with -L/... and -l it succeedsdoes not build with starpu/simgrid if symbols starpu_mpi_cache_flush and starpu_mpi_cache_flush_all_datarefix the problem of starpu mpi function detection - with the complete path of mpi libraries the test fails... with -L/... and -l it succeedsfix warning related to several main redefinition in the simgrid caseadapt chameleon to bu used with starpu+simgrid+mpiadapt Chameleon CMakeLists to be used with SIMGRID as a component of STARPU