rsz: Document wire splitting reason #5147
Conversation
On wire splitting in `repair_design`, name the inserted buffer according to the reason for splitting. A similar thing is done elsewhere in the repair code. Signed-off-by: Martin Povišer <[email protected]>
|
clang-tidy review says "All clean, LGTM! 👍" |
Signed-off-by: Martin Povišer <[email protected]>
|
clang-tidy review says "All clean, LGTM! 👍" |
|
The code changes are fine but you need to update the tests. |
Signed-off-by: Martin Povišer <[email protected]>
|
clang-tidy review says "All clean, LGTM! 👍" |
src/rsz/test/pin_swap1.ok
Outdated
| sky130_fd_sc_hd__nor4_2/C -> A B D | ||
| sky130_fd_sc_hd__nor4_2/A -> D B C | ||
| sky130_fd_sc_hd__nor4_2/B -> D A C | ||
| sky130_fd_sc_hd__nor4_2/C -> D A B |
There was a problem hiding this comment.
Huh, my local run and the CI server disagree on the iteration order here. Are the tests flaky?
There was a problem hiding this comment.
It looks like the perennial problem of stl sets comparing pointers:
typedef Set<LibertyPort*> LibertyPortSet;
in
sta::LibertyPortSet ports;
equivCellPins(cell, port, ports);
std::ostringstream ostr;
for (auto port : ports) {
ostr << ' ' << port->name();
}
Signed-off-by: Martin Povišer <[email protected]>
|
clang-tidy review says "All clean, LGTM! 👍" |
|
Now we are stuck on I guess the naming change influenced the iteration order somewhere downstream? How do I update those tests? |
|
save_flow_metrics |
|
Thanks! |
|
This means this PR will need a full CI testing as it may change results on all ORFS designs. Let me know when it is ready for that. |
On wire splitting in
repair_design, name the inserted buffer according to the reason for splitting. A similar thing is done elsewhere in the repair code.Split off from #5068