Skip to content

ATVA 2023
Compare
Choose a tag to compare
@SSoelvsten SSoelvsten released this 07 Jul 11:40
· 589 commits to main since this release
v4--ATVA2023
de2260d

Submission of our paper on Adiar 1.2.0 to ATVA 2023. To change the setting for this optimisation set adiar::memory_mode within the adiar_adapter's constructor. The possible values are:

  • adiar::memory_mode_t::EXTERNAL : Baseline with the performance of the previous algorithm.
  • adiar::access_mode_t::AUTO : Use internal memory, if the heuristic can guarantee it is safe to do so.
  • adiar::access_mode_t::INTERNAL : Always use internal memory

Some of the heuristics depend on changes to Adiar. To make use of these, change the submodule to the desired branch.

  • experiment/v1.2.1/only_node_cut: Disable any upper bounds derived from the 1-level cuts. Note, the computation for the cuts are still included.
  • experiment/v1.2.1/reduce_2level_cut: Run with the slightly slower, but more precise 2-level cut over-approximation.

Attached you can find the raw data and its analysis. These have also been uploaded to Zenodo.

Assets 9
Loading