merge OpenSTA 3.0 upstream/master into silimate/main#56
Conversation
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
…arallaxsw#370 Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
…arallaxsw#427) Signed-off-by: Drew Lewis <cannada@google.com>
…f fallback (parallaxsw#426) Signed-off-by: Drew Lewis <cannada@google.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
Signed-off-by: James Cherry <cherry@parallaxsw.com>
| deleteTagsPrev(); | ||
| arrivals_exist_ = true; | ||
| stats.report("Find clk arrivals"); | ||
| findAllArrivals(false, true); |
There was a problem hiding this comment.
This logic appears to have been replaced with findAllArrivals, but I'm unsure if I successfully propagated all the changes here.
| @@ -1229,7 +1167,8 @@ Search::computeClkGates() | |||
| VertexIterator vertex_iter(graph_); | |||
| while (vertex_iter.hasNext()) { | |||
| Vertex *vertex = vertex_iter.next(); | |||
| if (isClock(vertex)) { | |||
| TagGroup *tg = tagGroup(vertex); | |||
| if (tg && tg->hasClkTag()) { | |||
| libWarn(1199, line, "%s %s bus width mismatch.", type, "preset"); | ||
| preset_expr->deleteSubexprs(); | ||
| preset_expr = nullptr; | ||
| auto &gen_clk_groups = cell_group->findSubgroups("generated_clock"); |
There was a problem hiding this comment.
Generated clock parsing with the new API
| } | ||
|
|
||
| void | ||
| LibertyReader::visitFanoutLength(LibertyAttr *attr) | ||
| LibertyReader::makeTimingArcs(LibertyCell *cell, |
There was a problem hiding this comment.
This is another function I would appreciate a review on.
|
Hi @donn, I'll review the stuff that I implemented, which is around the generated clocks and clock gating. |
|
@stanminlee Would be immensely appreciated. |
| libWarn(1199, line, "%s %s bus width mismatch.", type, "preset"); | ||
| preset_expr->deleteSubexprs(); | ||
| preset_expr = nullptr; | ||
| auto &gen_clk_groups = cell_group->findSubgroups("generated_clock"); |
| Sta::clockGatedRegisters() | ||
| { | ||
| InstanceSeq Sta::clockGatedRegisters(const Mode *mode) { | ||
| Network* network = this->network(); |
There was a problem hiding this comment.
Why is this added? it is not used anywhere
There was a problem hiding this comment.
I am referring to network here in this comment
| InstanceSeq | ||
| Sta::clockGatedRegisters() | ||
| { | ||
| InstanceSeq Sta::clockGatedRegisters(const Mode *mode) { |
There was a problem hiding this comment.
Mode is also not used anywhere, why was it added?
| while (end_iter.hasNext()) { | ||
| auto end = end_iter.next(); | ||
| std::set<Delay> found_delays; | ||
| while (end_iter != end_iter_end) { |
There was a problem hiding this comment.
I have been running some tests, and there seems to be a bug here.
In the JSON output for the results, it sometimes prints a stray comma at the end of the checks like so:
[
{ ... }
{
last entry
},
]
which is invalid for JSON.
I think it has something to do with the end_iter == end_iter_end check, it may have some edge case with path endpoint deduplication.
I can look at this myself, but just letting you know since you may have more context around this.
9 participants
Most of this was just rote merge work, but some of it was more complex. Specifically, the LibertyReader was pretty much completely rewritten. That required a backport of Silimate's additions to all of:
…to the new system. This really should be reviewed by someone with more familiarity with these.