Skip to content

Clip: Improve performance and memory consumption

Spiros Tsalikis requested to merge spiros.tsalikis/vtk-m:clip-batching into master

The following set of improvements have been implemented for the Clip algorithm:

  1. Input points that are kept are determined by comparing their scalar value with the isovalue, instead of checking the output cells' connectivity.
  2. Output arrays are written only once, and they are not transformed. Due to that, no auxiliary arrays are needed to perform the transformations.
  3. A fast path for discarded and kept cells has been added, which are the most common cell cases.
  4. ClipTables are now more descriptive, and the non-inverted case tables have been imported from VTK, such that both VTK and VTK-m produce the same results.
  5. Employ batching of points and cells to use less memory and perform less and faster computations.

The new Clip algorithm:

  1. has up to 2.13x less memory footprint
  2. is up to 4.43x faster on the CPU
  3. is TODOx faster on the GPU
Edited by Spiros Tsalikis

Merge request reports

Loading