perf(main): reduce main sumcheck max degree from 4->3 - #1395
Closed
hero78119 wants to merge 65 commits into
Closed
Conversation
Train dynamic roots across the complete preflight workload and persist ABI-keyed native artifacts with checksum validation and atomic replacement. Track fallback reasons and recover safely through dynamic control flow and guarded memory accesses.
…t_aot_bb_workload # Conflicts: # ceno_emul/src/aot.rs # ceno_zkvm/src/e2e.rs
This reverts commit 14b2ad2.
This reverts commit 4f5dfed.
This reverts commit 5ebca74.
Admit only aligned LW/SW accesses sharing a base register and byte offset. Check the full affine interval for wrapping and region membership once, while retaining scalar guards for ungrouped or rejected accesses. Bump the cache identity because emitted code changes.
After block admission, keep the register base, dense-memory base, and memory ordinal base in their documented callee-saved registers. Emit memory events from the loaded cell metadata and defer next-PC/status stores until block exit. Scalar fallback, shard transitions, syscalls, and non-admitted entries continue to refresh context state.
Load generation, instance-count, and bucket-ceiling table bases once per admitted planner sequence. Preserve the existing generation and capacity checks, rollback path, and slow fallback for stale generations or bucket crossings.
Move public secp256k1 add/double normalization to halo2curves-axiom with Ceno-local variable-time field inversion. Preserve syscall witnesses, invalid inputs, infinity output, parity, and alias fallback; add deterministic differential and edge-case coverage.
Remove public cumulative preflight stages, tracking trainers, and generic compilation APIs. Production preflight now uses private scalar and admitted-block strategies, while tests exercise Pure value execution and FullTracer-compatible execution directly.
Provide a narrow production API that consumes the finalized preflight plan and replays every shard through FullTracer without assigning circuit witnesses. The replay preserves the exact preflight capacity and validates step, shard, exit-code, next-access, and boundary parity.
Gate AOT-only tracer, syscall, and benchmark support at their feature boundaries. Modernize deprecated secp byte conversions, make unsafe conversion types explicit, and document unavoidable transitive dependency-version lint exceptions so workspace all-target Clippy passes with warnings denied.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
Design Rationale
Change Highlights
Benchmark / Performance Impact
Operation
Layer
Benchmark command(s):
# paste exact command(s)Environment (CPU/GPU, core count, rust toolchain, commit hash):
raw data:
Testing
Risks and Rollout
Follow-ups (optional)
Copilot Reviewer Directive (keep this section)
When Copilot reviews this PR, apply
.github/copilot-instructions.mdstrictly.