brycewang-stanford HPCA Review Process
Use this to set expectations about how HPCA reviews a paper and decides its fate. Understanding the pipeline turns anxiety into a plan: you can predict what the rebuttal/revision window is for and what each outcome signals.
The pipeline
HPCA runs a double-blind review. After the late-July gate, papers are assigned to program-committee reviewers, reviews are written through the autumn, and — in recent cycles — authors get a single combined rebuttal/revision window: one period in which you both respond to reviews and may revise the paper, rather than two separate review rounds. Decisions follow a program-committee meeting; HPCA 2027's reported notification is November 6, 2026 (待核实 the exact rebuttal window).
What the single window means
Unlike venues with two review rounds, HPCA's combined window compresses "argue" and "change the paper" into one effort. Plan for both at once:
- Reviewers see your response and a revised paper, so a claim you defend in prose should also be reflected in the revision.
- The window is short; the objection ledger and staging shelf (see
hpca-author-responseandhpca-supplementary) must be ready before it opens.
Champion dynamics at the PC meeting
Architecture decisions are made by people in a room (or a call). A paper needs a champion — a reviewer willing to argue for it — and its fate often turns on whether the champion can answer the strongest objection. Write the rebuttal to arm your champion with the room, not only to satisfy the skeptic.
| Outcome | Likely meaning | Your move |
|---|---|---|
| Accept | The champion held; objections were addressable | Camera-ready + artifact evaluation |
| Reject with strong reviews | A fixable but unaddressed methodology or novelty gap | Fix, then retarget MICRO/ISCA/ASPLOS |
| Reject with weak/split reviews | No champion emerged, or scope mismatch | Reframe the contribution, re-route |
| Borderline / conditional | The room was divided | Read every review for the deciding objection |
Read reviews for the deciding objection
Not all objections weigh equally. Identify the one that decides the paper — usually a methodology or novelty concern the champion must overcome — and prioritize it over easy-to-fix surface comments. A rebuttal that answers three minor points and dodges the deciding one loses.
Confidentiality and preprints
Reviews and PC discussion are confidential; do not solicit or reveal reviewer identities, and do not de-anonymize yourself mid-cycle. Recent HPCA guidance asks PC members not to penalize arXiv/CAL preprints, so a preprint is allowed — but posting one that reveals authorship does not change the double-blind obligation on the submission itself.
Expectation-setting pass
[Stage] assigned / reviews out / window / PC decision / notified
[Window plan] response + revision both staged? (Y/N)
[Deciding objection] <the one that decides the paper>
[Champion arming] does the rebuttal give the champion the room? (Y/N)
[Outcome branches] accept / reject-strong / reject-split -> next action
Output format
[Pipeline position] <where the paper is>
[Rebuttal/revision window] <dates from the current CFP, or unknown>
[Priority objection] <one>
[Outcome-conditional plan] <accept / reject branches>
Reopen the current CFP for the exact review calendar and window dates — the pipeline timing is per-edition and was unconfirmed for 2027 at pack time.
Source: brycewang-stanford/Awesome-Journal-Skills → HPCA-Skills/skills/hpca-review-process/SKILL.md