5565848d91
There were issues with the particle advection code where a small number of work-heavy task invocations were needed. Since we were enforcing a minimum of 1024 invocations per thread, this effectively serialized scheduling. Now the scheduler dynamically adjusts for small thread launches, allowing finer scheduling. |
||
---|---|---|
.. | ||
ArrayManagerExecutionOpenMP.cxx | ||
ArrayManagerExecutionOpenMP.h | ||
AtomicInterfaceExecutionOpenMP.h | ||
CMakeLists.txt | ||
DeviceAdapterAlgorithmOpenMP.cxx | ||
DeviceAdapterAlgorithmOpenMP.h | ||
DeviceAdapterRuntimeDetectorOpenMP.cxx | ||
DeviceAdapterRuntimeDetectorOpenMP.h | ||
DeviceAdapterTagOpenMP.h | ||
ExecutionArrayInterfaceBasicOpenMP.cxx | ||
ExecutionArrayInterfaceBasicOpenMP.h | ||
FunctorsOpenMP.h | ||
ParallelQuickSortOpenMP.h | ||
ParallelRadixSortOpenMP.cxx | ||
ParallelRadixSortOpenMP.h | ||
ParallelScanOpenMP.h | ||
ParallelSortOpenMP.h | ||
VirtualObjectTransferOpenMP.h |