Skip to content

Tighten gf12 swerv_wrapper clocks - #4674

Merged
eder-matheus merged 2 commits into
The-OpenROAD-Project:masterfrom
The-OpenROAD-Project-staging:tight-swerv_wrapper-gf12
Oct 14, 2026
Merged

eder-matheus merged 2 commits into
The-OpenROAD-Project:masterfrom
The-OpenROAD-Project-staging:tight-swerv_wrapper-gf12

Conversation

@LucasYuki

Copy link
Copy Markdown
Contributor

Tighten gf12 swerv_wrapper clocks: 1500 ps → 1000 ps

File: flow/designs/gf12/swerv_wrapper/constraint.sdc

Constraint Before After
core_clock / vclk_core_clock period 1500 ps 1000 ps
jtag_clock / vclk_jtag_clock period 1500 ps 1000 ps
Core IO input/output delay 750 ps 500 ps (50% of period)
JTAG IO input/output delay 375 ps 250 ps (25% of period)

Why

  • Goal: a negative setup WNS. At 1500 ps the design meets setup with margin, and it keeps closing as the clock tightens (1400, 1300 and 1200 ps all met).
  • Both clocks are tightened together: jtag ↔ core paths are timed as synchronous, so mismatched periods create artificial edge offsets.
  • IO delays keep their original ratio to the period.

Results (full flow)

Period Setup WNS Setup TNS Setup viol. Hold WNS core_clock Fmax Route DRC
1500 ps +83.7 ps 0 0 −12.7 ps 706 MHz 0
1400 ps +42.6 ps 0 0 −10.0 ps 737 MHz 0
1300 ps +59.8 ps 0 0 −12.9 ps 806 MHz 0
1200 ps +45.4 ps 0 0 −10.0 ps 866 MHz 0
1000 ps −109.5 ps −44 404 ps 940 −10.9 ps 901 MHz 0

At 1000 ps the worst path is internal core logic (lsu_lsc_ctl.addr_in_pic_dc3ff → pic_ctrl_inst.picm_wre_flop). The design's limit is about 1110–1155 ps.

Signed-off-by: LucasYuki <lucasyuki@yahoo.com.br>
Signed-off-by: LucasYuki <lucasyuki@yahoo.com.br>
@LucasYuki LucasYuki self-assigned this Oct 13, 2026

@gemini-code-assist gemini-code-assist Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request updates the timing constraints in constraint.sdc for the swerv_wrapper design, reducing the clock periods for both the core clock and the JTAG clock from 1500 ps to 1000 ps, along with proportional adjustments to input and output delays. The review feedback correctly points out that treating the JTAG clock as synchronous to the core clock and tightening its period to 1 GHz is unrealistic and can negatively impact timing closure. The reviewer suggests keeping the JTAG clock at a realistic slower frequency (e.g., 100 MHz), defining the clock domains as asynchronous, and scaling the JTAG input/output delays accordingly.

@openroad-ci

Copy link
Copy Markdown
Member

🔍 QoR check

Metrics reflect the PR merge build — i.e. what will land on the target branch.

Commit 4e48805 · Jenkins build #1 · Baseline: build · View build on dashboard

62 design(s) checked — 0 with regression(s), 0 without a comparable baseline.

@eder-matheus
eder-matheus merged commit 08ddc64 into The-OpenROAD-Project:master Oct 14, 2026
12 checks passed
@eder-matheus
eder-matheus deleted the tight-swerv_wrapper-gf12 branch October 14, 2026 17:32
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants