Skip to content

generate_abstract: time a global-route abstract with the route's parasitics - #4688

Merged
maliberty merged 1 commit into
The-OpenROAD-Project:masterfrom
oharboe:abstract-grt-parasitics
Oct 17, 2026
Merged

maliberty merged 1 commit into
The-OpenROAD-Project:masterfrom
oharboe:abstract-grt-parasitics

Conversation

@oharboe

@oharboe oharboe commented Oct 16, 2026

Copy link
Copy Markdown
Collaborator

An abstract written from global route (ABSTRACT_SOURCE=5_1_grt) is now timed with global-route parasitics instead of placement parasitics, so its liberty matches the block's own timing at that stage.

XiangShan's MemBlock, one input-to-output arc (io_ooo_to_mem_intIssue_2_0_bits_lqIdx_value[0] → io_mem_to_ooo_wakeup_2_bits_pdest[7]):

arc (ps)
MemBlock's own STA at global route (-global_routing) 1,840
MemBlock's STA with -placement parasitics 6,911
abstract liberty before 6,753 – 6,915
abstract liberty after 1,844 – 2,035

In the parent, 196 of the 200 worst reg2reg paths launched through MemBlock's abstract; the worst, 7,855 ps, spent 6,912 ps on this arc.

  • generate_abstract: time a global-route abstract with the route's parasitics — from stage 5 on, without a SPEF, estimate_parasitics -global_routing, the same branch open.tcl takes.

No new test: a test that sees the change has to build a design through global route and compare its abstract with its own timing. The integration test is The-OpenROAD-Project/bazel-orfs#1281, which rebuilds every XiangShan abstract with it.

🤖 Generated with Claude Code

…sitics

generate_abstract read SPEF for a final-stage abstract and estimated
placement parasitics for every stage from place on, global route
included. An abstract written from 5_1_grt.odb (ABSTRACT_SOURCE=5_1_grt)
therefore modelled the block with placement parasitics, while the block
itself is timed at global route with the route's.

On a large block the two differ by a factor: XiangShan's MemBlock times
an input-to-output arc at 1,840 ps with global-route parasitics and
6,911 ps with placement parasitics, and the abstract's liberty carried
the 6.9 ns arc into the parent, where it set the parent's worst paths.

From global route on, without a SPEF, the abstract now estimates
parasitics from the global route, as open.tcl and global_route.tcl do.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>
@oharboe
oharboe marked this pull request as ready for review October 16, 2026 04:45

@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 generate_abstract.tcl script to estimate parasitics using global routing when the design stage is 5 or higher. The review feedback points out that if a SPEF file is already available at stage 5, reading it is more accurate than estimating parasitics. It suggests updating the logic to read the SPEF file for stage 5 if it exists, and only falling back to estimating parasitics if the file is missing.

Comment thread flow/scripts/generate_abstract.tcl Resolved
@oharboe
oharboe requested a review from maliberty October 16, 2026 04:45
@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 b309a78 · Jenkins build #1 · Baseline: build · View build on dashboard

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

@maliberty
maliberty merged commit 7bcaf47 into The-OpenROAD-Project:master Oct 17, 2026
12 checks passed
oharboe added a commit to oharboe/OpenROAD-flow-scripts that referenced this pull request Oct 20, 2026
generate_abstract.tcl conflicted with The-OpenROAD-Project#4688 (695436e), which this
change supersedes: the flagless estimate_parasitics times an abstract
from global route on with the route's parasitics, and the SPEF is read
only where final_extracts_parasitics says one was made. Kept this
branch's side.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: Øyvind Harboe <oyvind.harboe@zylin.com>
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