Tage trace fix

[CJ362ff]

Summary by CodeRabbit

• Changes
  • Modified branch prediction update behavior by disabling "update on read" functionality by default
  • Corrected prediction trace recording to capture original prediction data instead of recomputed values for improved accuracy

[coderabbitai]

📝 Walkthrough

Walkthrough

This PR disables the default "update on read" behavior in the BTBTAGE branch predictor configuration and refactors trace recording logic in the update function to source prediction metadata fields from the original prediction snapshot instead of recomputed values.

Changes

Cohort / File(s)	Summary
Configuration Parameter src/cpu/pred/BranchPredictor.py	Changed updateOnRead parameter default from True to False in BTBTAGE class configuration.
Trace Recording Logic src/cpu/pred/btb/btb_tage.cc	Refactored BTBTAGE::update() to source trace fields (main_info, alt_info, useAlt, taken) from original prediction snapshot (trace_pred) rather than recomputed prediction.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Possibly related PRs

• Restore/per table numways #798: Overlaps with refactoring of BTBTAGE::update() to source prediction fields from predMeta->preds instead of recomputed values.
• Microtage merage  #775: Related through use of per-entry TageMeta preds and TAGE prediction metadata structures.

Suggested labels

align-kmhv3

Suggested reviewers

• jensen-yan

Poem

🐰 The BTB TAGE hops along with grace,
No read-updates cluttering the place!
Traces now peek at the original sight,
Not recomputed dreams taking flight,
Branch predictions aligned just right! ✨

🚥 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 'Tage trace fix' directly relates to the main changes in the pull request, which involve fixing trace record derivation in BTBTAGE::update method and adjusting the updateOnRead parameter.
Docstring Coverage	✅ Passed	Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.

_{✏️ 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 tage-trace-fix

---

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

• X
• Mastodon
• Reddit
• LinkedIn

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

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@src/cpu/pred/btb/btb_tage.cc`:
- Around line 798-806: The code currently reads predMeta->preds[btb_entry.pc]
which default-inserts a trace prediction when the key is missing, mutating
predMeta and producing incorrect trace fields; replace that lookup with the
already-resolved source (use original_pred if present or recomputed if not)
instead of indexing predMeta->preds — i.e., obtain trace_pred from the existing
original_pred/recomputed variable used earlier and pass its fields into t.set
(keeping the same main_info/alt_info/taken/useAlt/alloc_success usage) so you
avoid implicit insertion into predMeta->preds.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 5c0f1254-dc45-44dd-b24d-226aba928c4a

📥 Commits

Reviewing files that changed from the base of the PR and between abf85ff and 3e02cc5.

📒 Files selected for processing (2)

• src/cpu/pred/BranchPredictor.py
• src/cpu/pred/btb/btb_tage.cc

[github-actions]

🚀 Coremark Smoke Test Results

Branch	IPC	Change
Base (xs-dev)	2.2709	-
This PR	2.2713	📈 +0.0004 (+0.02%)

✅ Difftest smoke test passed!