Skip to content

Tage way tunning and enlarge L3 size to 32MB#811

Closed
edwu186 wants to merge 2 commits intoxs-devfrom
regression-L3-32MB
Closed

Tage way tunning and enlarge L3 size to 32MB#811
edwu186 wants to merge 2 commits intoxs-devfrom
regression-L3-32MB

Conversation

@edwu186
Copy link
Copy Markdown
Contributor

@edwu186 edwu186 commented Apr 3, 2026

tage way tunning and enlarge L3 size to 32MB

Summary by CodeRabbit

  • Configuration Updates
    • L3 cache size now explicitly set to 32MB in default system configuration.
    • Branch predictor table parameters optimized with adjusted capacity allocation and associativity for improved cache performance.

edwu186 added 2 commits March 27, 2026 19:03
Change-Id: Ibe23fc57b667b597292eec6847b4760242aa372d
Change-Id: Ie3e72b9396ad0ca0a25d332ef2b1b4cffcac0c5a
@coderabbitai
Copy link
Copy Markdown

coderabbitai bot commented Apr 3, 2026

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info
⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: a4caa6f6-2e28-461e-9c09-6e861ac90536

📥 Commits

Reviewing files that changed from the base of the PR and between 4b6b1dc and 5c78e43.

📒 Files selected for processing (2)
  • configs/example/idealkmhv3.py
  • src/cpu/pred/BranchPredictor.py

📝 Walkthrough

Walkthrough

Configuration parameters for the KMH v3 system are enhanced: L3 cache size is now explicitly set to 32MB in the ideal configuration, and BTBTAGE branch predictor table geometry is tuned by increasing table T2–T6 capacities to 8192 entries and raising table T2 associativity to 4 ways.

Changes

Cohort / File(s) Summary
Cache Configuration
configs/example/idealkmhv3.py
L3 cache size explicitly set to 32MB in default parameters.
Branch Predictor Tuning
src/cpu/pred/BranchPredictor.py
BTBTAGE table geometry adjusted: tables T2–T6 increased to 8192 entries, table T2 raised to 4-way associativity.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~5 minutes

Possibly related PRs

Suggested labels

perf, align-kmhv3

Suggested reviewers

  • jensen-yan
  • tastynoob

Poem

🐰 A tweak here, a tweak there,
L3 caches with such care,
TAGE tables grow so wide and deep,
Better predictions our CPU shall reap!
Performance hops along the way... 🌟

🚥 Pre-merge checks | ✅ 3
✅ Passed checks (3 passed)
Check name Status Explanation
Description Check ✅ Passed Check skipped - CodeRabbit’s high-level summary is enabled.
Title check ✅ Passed The title accurately reflects the main changes: TAGE way tuning and L3 size enlargement to 32MB are both present in the changeset.
Docstring Coverage ✅ Passed No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

✏️ Tip: You can configure your own custom pre-merge checks in the settings.

✨ Finishing Touches
📝 Generate docstrings
  • Create stacked PR
  • Commit on current branch
🧪 Generate unit tests (beta)
  • Create PR with unit tests
  • Commit unit tests in branch regression-L3-32MB

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

Comment @coderabbitai help to get the list of available commands and usage tips.

@edwu186 edwu186 requested a review from jensen-yan April 3, 2026 03:59
updateOnRead = Param.Bool(True, "Enable update on read, no need to save tage meta in FTQ")
numPredictors = Param.Unsigned(8, "Number of TAGE predictors")
tableSizes = VectorParam.Unsigned([2048, 2048, 2048, 2048, 2048, 2048, 2048,2048],"the TAGE T0~Tn length")
tableSizes = VectorParam.Unsigned([2048, 2048, 8192, 8192, 8192, 8192, 8192, 2048], "the TAGE T0~Tn length")
Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

不要在这里修改tage 配置,应该在idealkmhv3.py 中修改,这里修改会让kmhv3.py 配置也被拉大了

@github-actions
Copy link
Copy Markdown

github-actions bot commented Apr 3, 2026

🚀 Coremark Smoke Test Results

Branch IPC Change
Base (xs-dev) 2.2665 -
This PR 2.3043 📈 +0.0378 (+1.67%)

✅ Difftest smoke test passed!

@edwu186 edwu186 closed this Apr 3, 2026
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.

2 participants