Skip to content

[WIP] Enhance MTLN performance#401

Draft
lmdiazangulo wants to merge 5 commits into
devfrom
400-mtln-cpu-performance-improvements
Draft

[WIP] Enhance MTLN performance#401
lmdiazangulo wants to merge 5 commits into
devfrom
400-mtln-cpu-performance-improvements

Conversation

@lmdiazangulo
Copy link
Copy Markdown
Contributor

@lmdiazangulo lmdiazangulo commented May 4, 2026

Profiling seems to show that the problem is with ngspice. A possible solution would be to keep calls to it at minimum. Just when the input contains "circuits" which require it.

@lmdiazangulo lmdiazangulo changed the title [WIP] Enhance MTLN performance by optimizing bundle processing and reducing… [WIP] Enhance MTLN performance May 12, 2026
- Add termination_handler.F90: Fortran-based solvers for simple RLC terminations
  (types 1-10: SHORT, OPEN, SERIES, PARALLEL, and complex RLC combinations)
- Add benchmark.F90: Timing utilities for performance measurement
- Modify network.F90: Add termination type and component values to nw_node_t
- Modify network_manager.F90: Implement hybrid approach using Fortran for simple
  terminations and ngspice only for complex ones (CIRCUIT, NETWORK)
- Modify preprocess.F90: Populate termination info when creating network nodes
- All 31 MTLN tests pass with the new implementation
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant