Skip to content
Snippets Groups Projects

Repository graph

You can move around the graph by using the arrow keys.
Select Git revision
  • add-py3-conda-ci
  • array-svm
  • bitonic-sort
  • cffi
  • cffi-c++0x
  • circumvent-nvidia-enqueue_fill_buffer-bug
  • concurrency-check
  • deprecated-boost-python
  • direct-to-pocl
  • examples-ci
  • fix-info-type-memory-leak
  • fix-ro-svm-crashes
  • gpl-special-functions
  • intel-bitonic-size0-xfail
  • intel-cl-debug
  • intel-gpu-fixes
  • itruediv
  • main default protected
  • modernize-ci
  • more-pyproject
  • v2025.1
  • v2024.3
  • v2024.2.7
  • v2024.2.6
  • v2024.2.5
  • v2024.2.4
  • v2024.2.3
  • v2024.2.2
  • v2024.2.1
  • v2024.2
  • v2024.1
  • v2023.1.4
  • v2023.1.3
  • v2023.1.2
  • v2023.1.1
  • v2023.1
  • v2022.3.1
  • v2022.3
  • v2022.2.4
  • v2022.2.3
40 results
Created with Raphaël 2.2.019Jan18171587431Dec19872127Nov21222119181716118131Oct2925191412728Sep2322171615141311109831Aug3023202120171615915Jul14131229Jun27262523222019181617161514429May28242312111029Apr23231Mar3029121129Jan282795Dec425Nov1715141276565328Oct272623222119Sep161529Aug27261198327Jul261914May330Apr29519Mar18151415141312987423Feb14131213121312631Jan282726272625262423317Dec627Nov2625201413765128Oct27231187527Sep25232119181716141311121021Aug20191716Pack indices and args into the same tuple for set_arg*multiIntroduce _set_arg{,_buf}_multi for batched argument settingMake creation of KernelWithCustomEnqueue easier to read (thanks @kaushikcfd)Wrapper: switch some C++ args from py::object to py::handleInvoker generator: globalize some py_codegen importsRemove unused generate_buffer_arg_setterUse positional args in _cl.enqueue_nd_range_kernel in invoker to avoid pybind slowness with kwargsRevert "Rewrite first_arg_dependent_memoize without decorator package, drop dep"Rewrite first_arg_dependent_memoize without decorator package, drop depkernel_get_work_group_info: Use Device.int_ptr in cache keyTurn off generation of debug helper code in invoker with sys.flag.optimizeRename KernelWithOverriddenCall->KernelWithCustomEnqueueStreamline invoker code (and its generation)Fix overriding of custom set_args in KernelWithOverriddenCallDo not cache result in _get_common_dtype if obj2 is of type intMake a kernel-specific class to override __call__ and avoid an indirect call on kernel enqueueInvoker gen: check for _KERNEL_ARG_CLASSES before NoneUpdate compyte for equal_strides fast pathDo not recompute flags in arrays made by _new_like_meAvoid generator expression in checking for negative shape entriesCache result of _get_common_dtypeInline _make_strides in pyopencl.arrayRemove a pypy compatibility hack for checking for object-array-nessRevert "Create special-case set_arg wrappers for common integer/float types"Actually pass a device to _splay on invocations outside pyopencl.arrayAvoid a few calls to len() in enqueue_nd_range_kernelFix a few remaining invocations of splayPrivatize cl.array.{splay,Array.get_sizes}, cache _get_sizes resultsAvoid malloc in enqueue_nd_range_kernelBump versionCreate special-case set_arg wrappers for common integer/float typesMerge pull request #415 from inducer/array-kernel-exec-hookAdd a note that ARRAY_KERNEL_EXEC_HOOK is undocumentedIgnore not-callable pylint warning on ARRAY_KERNEL_EXEC_HOOKAdd ARRAY_KERNEL_EXEC_HOOKMerge branch 'inducer-master-patch-80983' into 'master'Set autoclass_content = "class", autodoc_typehints = "description", remove irrelevant stuff from doc/confMerge pull request #414 from inducer/fix-rng-evt-mgmtRemove another now-unnecessary add_event in clrandom.randFix .event -> .events typo