Repository navigation
update OR submodule - #4779
update OR submodule#4779maliberty merged 1 commit into
Conversation
Signed-off-by: Matt Liberty <mliberty@precisioninno.com>
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
🔍 QoR checkMetrics reflect the PR merge build — i.e. what will land on the target branch. Commit 62 design(s) checked — 12 with regression(s), 1 without a comparable baseline. ❌ asap7/aes_lvt base — 1 failing metric(s)
❌ asap7/riscv32i base — 1 failing metric(s)
❌ asap7/riscv32i-mock-sram base — 1 failing metric(s)
❌ ihp-sg13g2/gcd base — 3 failing metric(s)
❌ ihp-sg13g2/ibex base — 1 failing metric(s)
❌ ihp-sg13g2/jpeg base — 1 failing metric(s)
❌ nangate45/tinyRocket base — 3 failing metric(s)
❌ sky130hd/chameleon base — 5 failing metric(s)
❌ sky130hd/jpeg base — 3 failing metric(s)
❌ sky130hd/microwatt base — 1 failing metric(s)
❌ sky130hs/jpeg base — 2 failing metric(s)
❌ sky130hs/riscv32i base — 3 failing metric(s)
|
|
@maliberty Merge? |
|
I am debugging the QoR regressions in the designs to investigate before this is merged. aes_lvt, riscv32i, riscv32i-mock-sram (includes the riscv32i config), ihp gcd, ihp ibex, nangate45 tinyRocket, sky130hd chameleon, sky130hd jpeg, sky130hs jpeg, sky130hs riscv32i. The other two failures are tiny antenna-diode count changes (ihp jpeg 100→102, microwatt 2117→2150). Those designs don't use PARALLEL_EVAL and are probably just noise from the placement and routing changes. TNS already gets worse at CTS in several designs (sky130hs/riscv32i goes from −1.0 to −50.3). ORFS runs repair_timing_helper in cts.tcl, so the damage happens in setup repair. Suspect OR commits in the bump (98c6045c72 → f3d27840b) These two only affect PARALLEL_EVAL runs:
AI ruled these out:
|
|
@dsengupta0628 @maliberty Can you debug this after the merge? Its going to take more updates to OpenROAD anyway. That should get us unstuck. |
|
I debugged ihp-sg13g2/gcd as this was the smallest. The regression comes from #11974 (1740e32a2e, "rsz: Scale the PARALLEL_EVAL WNS thresholds by the clock period"). The commit changes the floorplan-stage result slightly, and that difference grows through placement. Adding @jhkim-pii Are there changes in your pipeline to address the QoR changes outlined here (is this going to help The-OpenROAD-Project/OpenROAD#12027 for example)? The open PARALLEL_EVAL PRs (#12027 and others) will move these numbers again, so my be worth merging and recalibrating the baselines after those PRs land? I cannot override to merge this PR unfortunately. |
No description provided.