Our HDF5 backend does not yet implement proper chunking. Note: we should stay with *fixed size* data spaces nonetheless for performance. References: - https://support.hdfgroup.org/HDF5/doc/Advanced/Chunking/ - https://support.hdfgroup.org/HDF5/faq/parallel.html Pitfalls: - https://github.com/ComputationalRadiationPhysics/libSplash/issues/133 - https://github.com/ComputationalRadiationPhysics/picongpu/issues/2841 libSplash implementation: - https://github.com/ComputationalRadiationPhysics/libSplash/blob/master/src/DCDataSet.cpp#L169-L170 - https://github.com/ComputationalRadiationPhysics/libSplash/blob/master/src/include/splash/core/DCHelper.hpp#L87-L104
Our HDF5 backend does not yet implement proper chunking.
Note: we should stay with fixed size data spaces nonetheless for performance.
References:
Pitfalls:
libSplash implementation: