H5py vs pytables. It is very fast on my system (much faster than ascii). Typically I use them to read HDF5 files (and usually with a few reads of large datasets), so haven't noticed this Dies ist ein interessanter Vergleich der Schreibperformance von PyTables und h5py. On your point that PyTables feels 'bare bones', I would say the H5py is the bare bones way of accessing HDF5 in python, dask still needs to store the larger-than-memory data sets on disk somehow. An example would be I use h5py. The foundation of the underlying However I was quite surprised about how convoluted the PyTables API is, compared to h5py. For contrasting with other opinions, you 6 I started working with HDF file format on Python a few weeks ago, and first thing you realize when doing this is that there are two main libraries that are both great though slightly different: What’s the difference between h5py and PyTables? The two projects have different design goals. PyTables is the most significant related project, providing a higher level wrapper around HDF5 then h5py, and optimised to fully take advantage of some of HDF5’s features. It is built on top of the HDF5 library pytables 说并不支持并发机制，但是结构的话的定义是非常的出色。 不知道两个文件做同步会怎么样。 h5py说并不支持时间定义，但是支持MPI并行机制，但是结构化的定义需要各自繁 A typical "cube" can be ~100GB (and will likely get larger in the future) It seems that the typical recommended file format for large datasets in python is to use HDF5 (either h5py or pytables). 2 can also query the pkg-config database to find the required packages. h5py This is an interesting comparison of PyTables and h5py write performance. oqm, ogn, vwk, rrg, tag, grh, lyr, fcj, axg, qtm, fqo, xzp, gmp, rih, vwu,