IPComp is an interpolation-prediction-based lossy compressor designed specifically for scientific datasets. Its key feature is progressive compression, enabling multiple levels of decompression precision from a single compressed file. This means that higher-fidelity data can be reconstructed incrementally, requiring only additional data to be loaded rather than reloading the entire dataset.
Even in non-progressive compression scenarios, where a single fixed precision is required, IPComp achieves high compression ratios, making it an efficient choice for lossy data reduction.
- Progressive Decompression: A single compressed file supports multiple precision levels, reducing storage and computation overhead for adaptive applications.
- Incremental Loading: Higher-fidelity reconstructions require only additional data, minimizing I/O costs.
- High Compression Efficiency: Even in standard, non-progressive compression use cases, IPComp achieves competitive compression ratios.
Command-line Syntax:
IPComp <datafile> -dataType[f/d] -dim_num <dim1> <dim2> ... -bound_mode -bound_num <bound1> <bound2> ...
Example Commands:
- Error-Bounded Progressive Compression
./src/IPComp density.d64 -d -3 256 384 384 -error -3 1e-3 1e-4 1e-5
This compresses the dataset density.d64 with progressive error bounds at 1e-3, 1e-4, and 1e-5.
- Bitrate-Bounded Progressive Compression
./src/IPComp density.d64 -d -3 256 384 384 -bitrate -3 1.0 2.0 3.0
This compresses density.d64 using bitrate constraints of 1.0, 2.0 and 3.0 bits per value.
IPComp is licensed under the BSD 3-Clause License.
This software is based on SZ (Version 3.0), originally developed by Argonne National Laboratory.
Original SZ Authors: Sheng Di, Kai Zhao, Xin Liang, Dingwen Tao, Franck Cappello
Copyright © 2016, UChicago Argonne, LLC
mkdir build && cd build
cmake ..
make
Then, you'll find all the executables in [INSTALL_DIR]/test