Reduce the compile time for quadrature representation (-O)
Reduce the compile time when using the optimisations for quadrature representation. It might be possible to improve the performance if we move some parts to C++ and perhaps use Swig to generate the module.
Write about optimisations in manual.
Blueprint information
- Status:
- Complete
- Approver:
- None
- Priority:
- Essential
- Drafter:
- None
- Direction:
- Needs approval
- Assignee:
- None
- Definition:
- Approved
- Series goal:
- None
- Implementation:
- Implemented
- Milestone target:
- 0.7.1
- Started by
- Garth Wells
- Completed by
- Garth Wells
Related branches
Related bugs
Sprints
Whiteboard
Also need to reduce memory usage.
Can we close this one? We've implemented some new and lighter optimisations and the freedom to choose between them.
The compile time can of course always be improved, but I think we've come a long way since this blueprint was registered.
I think that Swig is a dead end for this purpose, and it will not help reduce the memory usage.
To do this we'll need to figure out how to free memory while generating the code, this will require quite some work and I think it should be in a new blueprint. If we are able to do this, one could use -r quadrature -O -f simplify_expression on a _really_ complex form and just let it compile over night.