In the next branch, we now have an optimization that improved ckd_mul performance, but there is another problem on ckd_add contributing to the regression: [bugs:#3602].
In the next branch, the implementation of chk_add has been changed so that all calls get inlined; code size should be back down to normal levels now.
But there is still a bit more potential for further optimization by using cheaper types. Either manually, or automatically as an optimization based on information from generalized constant propagation (which is currently being developed in the genconstprop branch).
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
It's the strtoul fix in [r13811]. Nothing to be done about it now. The way forward here is a more efficient ckd_mul.
Related
Commit: [r13811]
Ticket moved from /p/sdcc/bugs/3562/
Can't be converted:
In the next branch, we now have an optimization that improved ckd_mul performance, but there is another problem on ckd_add contributing to the regression: [bugs:#3602].
Related
Feature Requests: #880
Last edit: Philipp Klaus Krause 2023-06-11
In the next branch, the implementation of chk_add has been changed so that all calls get inlined; code size should be back down to normal levels now.
But there is still a bit more potential for further optimization by using cheaper types. Either manually, or automatically as an optimization based on information from generalized constant propagation (which is currently being developed in the genconstprop branch).
Fixed by merging the next and genconstprop branches to trunk a while ago.