Changes between Initial Version and Version 1 of Ticket #4879, comment 7


Ignore:
Timestamp:
2018-04-12T08:01:49Z (7 years ago)
Author:
Martin Sjölund

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #4879, comment 7

    initial v1  
    22> Replying to [comment:1 sjoelund.se]:
    33> But we could and probably should mark all functions in {{{<modelName>_06inz.c}}} and {{{<modelName>_08bnd.c}}} with the non-optimization flags, since the non-linear parts are in the file {{{<modelName>_02nls.c}}}.
    4 > > 
     4
     5I am not fully certain. Most bindings are marked I believe. Better gains would come from merging identical equations from continuous-time and simulation-time (since those equations are already optimized).
     6
    57> > > It is also advisable to optimize the C code, e.g. avoiding unnecessary macro expansions (see #4871), and moving all structural information (incidence matrices and similar) from the C sources to XML files.
    68> >