This video describes the differences between Tecplot native file types: DAT, PLT and SZPLT, and when using each type is the most beneficial.
In this video, the uses and differences between Tecplot 360’s three different native data file types are described.
DAT: The first and most basic data type is the ASCII DAT data format (*.dat). This is a simple human-readable or -writeable format and is perfect for smaller data sets when working between multiple applications. However, when data sets become large, the inefficiency of ASCII data storage quickly becomes apparent and a binary format should be considered.
PLT: The second data type is the Tecplot PLT binary format (*.plt). This format is Tecplot’s traditional format and is commonly exported by solvers. PLT format is significantly more efficient than the plain text ASCII format.
SZPLT: The third data format is SZL or Sub-Zone Load-on-demand (*.szplt). This format subdivides and catalogs volume zones so that when working with the data set, it will only load data from the needed areas. These techniques make the SZL format the most efficient and significantly faster than the traditional PLT format.
In a simple test of placing a single slice in the volume of a 7.1 million element case, you can see how each data file format performs.
- Data file size: binary SZPLT and PLT have 70% and 50% compression ratios respectively over the ASCII file format.
- Load times: ASCII file loads in just over 5 minutes while the two binary SZPLT and PLT formats load almost instantly. Clearly, ASCII format is not a good choice for data of this size. SZL really shines when grids are 100 million elements and more, where load times can beat PLT by an order of magnitude.
- Memory consumption: The final metric that differentiates these file types is memory consumption. ASCII uses a significant amount of memory where SZL uses about 1/3 as much memory as PLT. As datasets get larger the memory benefits of SZL continue to outpace PLT as SZL loads only a fraction of the total volume.
Thanks for watching!