Skip to main content

4 posts tagged with "release"

View All Tags

· One min read
Flavio Lorez

Fixes for paralell writing. Writing is sped up significantly.

Test for writing speed​

warning

out_directory is where a database will be created. Make sure it is an empty directory.

  • Write 1000 steps with 1, 2, 4 and 8 threads. Timing is printed to stdout. 
    ./tests/paralellization/test_steps/run.sh out_directory
  • Write single big array (20'000x20'000) with 1, 2, 4 and 8 threads. Timing is printed to stdout. 
    ./tests/paralellization/test_big_array/run.sh out_directory

· One min read
Flavio Lorez

Added functionality for userdata. To show the content of it, display the object sim.userdata.

Simple way to store non field data​

  • Scalar, string, etc will are stored as attributes: sim.userdata['some_name'] = value
  • Arrays are stored as datasets: sim.userdata['array_name'] = arr

Access is natural: sim.userdata['some_name'] -> returns the stored value

· 2 min read
Flavio Lorez

Introduced globally unique identifiers for each database. They are stored as a file in the database directory. This allows to safely link different simulations together, e.g. meshes from a mesh database that are used in many places. Also, it allows to access any database from any path without remembering where it is stored. To do so, an index is maintained at ~.config/bamboost which contains all previously accessed databases.

Also introduced nice reprs for a database and for a simulation inside jupyter notebooks 😄

In addition, some internal things were optimized (don't remember what exactly). Hopefully, nothing was broken. Testing is still to do.

· One min read
Flavio Lorez

Initial post​

This is the initial post of the blog. It is a test to see if the blog is working properly.