update failing metrics - #4452
Conversation
Signed-off-by: Eder Monteiro <emrmonteiro@precisioninno.com>
Signed-off-by: Eder Monteiro <emrmonteiro@precisioninno.com>
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request updates the tools/OpenROAD submodule commit and adjusts the baseline rules and thresholds for the gf12/ca53 and sky130hd/microwatt designs. The review feedback highlights critical regressions introduced by the OpenROAD update: a severe timing regression in gf12/ca53 (with setup WS and TNS degrading significantly) and a substantial increase in the global routing antenna diode count for sky130hd/microwatt. It is recommended to investigate the root causes of these regressions in the tool rather than relaxing the baseline rules to accommodate them.
| "value": -900.0, | ||
| "compare": ">=" | ||
| }, | ||
| "cts__timing__setup__tns": { | ||
| "value": -2080.0, | ||
| "value": -6700.0, |
There was a problem hiding this comment.
The timing metrics for gf12/ca53 show a severe regression:\n- cts__timing__setup__ws degraded from -248.0 to -900.0 (a ~260% degradation).\n- cts__timing__setup__tns degraded from -2080.0 to -6700.0 (a ~220% degradation).\n\nRelaxing the regression limits to accommodate such a massive timing regression is highly discouraged as it masks potential tool or flow bugs introduced by the OpenROAD submodule update. The root cause of this timing regression in the tool should be investigated and resolved instead of updating the baseline rules to accept these degraded values.
| }, | ||
| "globalroute__antenna_diodes_count": { | ||
| "value": 2520, | ||
| "value": 3427, |
There was a problem hiding this comment.
The antenna diode count for sky130hd/microwatt during global routing has increased significantly by ~36% (from 2520 to 3427). This substantial increase in diode insertion can lead to routing congestion and area overhead. This indicates a potential regression or change in the antenna avoidance/diode insertion behavior of the updated OpenROAD tool that should be investigated.
maliberty
left a comment
maliberty
left a comment
There was a problem hiding this comment.
ca53 needs more investigation.
🔍 QoR checkMetrics reflect the PR merge build — i.e. what will land on the target branch. Advisory — results are log-only and do not affect build status. The authoritative QoR gate remains the local rules-file check. Commit 62 design(s) checked — 3 with regression(s), 0 without a comparable baseline.
❌ asap7/ethmac base — 2 failing metric(s)
❌ nangate45/mempool_group base — 2 failing metric(s)
❌ sky130hd/microwatt base — 2 failing metric(s)
|
3 participants
No description provided.