feat: Replace converging jmpif with empty block destinations with a single jmp by vezenovm · Pull Request #8613 · noir-lang/noir

vezenovm · 2025-05-21T21:17:28Z

Description

Problem*

No issue but something I noticed originally when making the pruning dead parameters PR and was surprised to not see certain jumps simplified.

Summary*

Attempt to simplify a jmpif terminator if both branches converge.

We define convergence as when two branches of a jmpif ultimately lead to the same
destination block, after following chains of empty blocks. If they do, the conditional
jump is unnecessary and can be replaced with a simple jmp.

There is room for follow-up PRs that do not restrict jump chains to empty blocks. For example, if we had entirely pure instructions in a block it would also be safe to follow the jump chain as we do in this PR. For now, we restrict a jump chain to only be empty blocks with no parameters.

Additional Context

Documentation*

Check one:

No documentation needed.
Documentation included in this PR.
[For Experimental Features] Documentation to be submitted in a separate PR.

PR Checklist*

I have tested the changes locally.
I have formatted the changes with Prettier and/or cargo fmt on default settings.

…to mv/simplify-converging-jmpif

github-actions · 2025-05-21T21:27:40Z

Changes to Brillig bytecode sizes

Generated at commit: 1590ada225d3c5195fd73e8e71e5b9251f0336d5, compared to commit: 9b7320908e74377bfd55a8760f16974ef1d07c05

🧾 Summary (10% most significant diffs)

Program	Brillig opcodes (+/-)	%
to_le_bytes_inliner_max	-2 ✅	-1.72%
to_le_bytes_inliner_zero	-2 ✅	-1.72%

Full diff report 👇

Program	Brillig opcodes (+/-)	%
slices_inliner_min	2,163 (-3)	-0.14%
slices_inliner_max	1,693 (-3)	-0.18%
slices_inliner_zero	1,625 (-3)	-0.18%
to_le_bytes_inliner_min	123 (-2)	-1.60%
to_le_bytes_inliner_max	114 (-2)	-1.72%
to_le_bytes_inliner_zero	114 (-2)	-1.72%

github-actions · 2025-05-21T21:27:40Z

Changes to number of Brillig opcodes executed

Generated at commit: 1590ada225d3c5195fd73e8e71e5b9251f0336d5, compared to commit: 9b7320908e74377bfd55a8760f16974ef1d07c05

🧾 Summary (10% most significant diffs)

Program	Brillig opcodes (+/-)	%
slices_inliner_zero	-3 ✅	-0.11%
slices_inliner_max	-3 ✅	-0.12%

Full diff report 👇

Program	Brillig opcodes (+/-)	%
slices_inliner_min	3,802 (-3)	-0.08%
to_le_bytes_inliner_min	1,016 (-1)	-0.10%
to_le_bytes_inliner_max	1,003 (-1)	-0.10%
to_le_bytes_inliner_zero	1,003 (-1)	-0.10%
slices_inliner_zero	2,736 (-3)	-0.11%
slices_inliner_max	2,493 (-3)	-0.12%

TomAFrench · 2025-05-21T21:29:48Z

Be aware that this can break flattening (I think).

TomAFrench · 2025-05-21T21:31:28Z

~~Ah, I misunderstood what you meant by converging~~ nvm, misread as I was on mobile. This is the case that can break things.

michaeljklein

Small doc change but LGTM

compiler/noirc_evaluator/src/ssa/opt/simplify_cfg.rs

Co-authored-by: Michael J Klein <michaeljklein@users.noreply.github.com>

github-actions

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'Execution Time'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.20.

Benchmark suite	Current: `ff10b51`	Previous: `9b73209`	Ratio
`semaphore-depth-10`	`0.024` s	`0.019` s	`1.26`

This comment was automatically generated by workflow using github-action-benchmark.

CC: @TomAFrench

…to mv/simplify-converging-jmpif

…ingle jmp (#8613) Co-authored-by: Michael J Klein <michaeljklein@users.noreply.github.com>

Automated pull of nightly from the [noir](https://github.com/noir-lang/noir) programming language, a dependency of Aztec. BEGIN_COMMIT_OVERRIDE fix(fmt): correctly format mixed secondary attributes and doc comments (noir-lang/noir#8735) fix!: require types for trait impl associated constants (noir-lang/noir#8734) fix!: Prevent returning references from if expressions (noir-lang/noir#8731) fix: cast signed to u1 follow-up (noir-lang/noir#8730) fix: cast signed to u1 (noir-lang/noir#8720) fix: do not mutate arrays later copied inside other arrays (noir-lang/noir#8701) chore: box `AsTraitPath` and `TypePath` in `ExpressionKind` (noir-lang/noir#8716) fix: general solution for accessing associated constants (noir-lang/noir#8417) fix(ssa): Validate checked signed add/sub is followed by a truncate (noir-lang/noir#8706) chore: add pre- and post-check on `array_set` optimization pass (noir-lang/noir#8714) fix: merge expr bindings with instantiations bindings during monomorphization (noir-lang/noir#8713) fix: better way to do LSP file overrides (noir-lang/noir#8702) fix(fmt): correct indentation when formatting long struct patterns (noir-lang/noir#8711) fix(fuzz): Prevent breaking/continuing out from let blocks in the AST fuzzer (noir-lang/noir#8708) chore: remove override for zero length inputs (noir-lang/noir#8709) feat: Replace converging jmpif with empty block destinations with a single jmp (noir-lang/noir#8613) feat(cli): Add `nargo interpret` command (noir-lang/noir#8700) chore: more 1-tuple printing fixes (noir-lang/noir#8699) chore(fuzz): Fuzz the SSA parser (noir-lang/noir#8647) fix: (SSA parser) translate blocks in logical order rather than syntax order (noir-lang/noir#8668) fix(SSA): show and parse range_check's assert_message (noir-lang/noir#8652) chore: Show the step number in the SSA message (noir-lang/noir#8698) chore(docs): Add pointers to tests (noir-lang/noir#8695) chore(fuzz): Capture comptime print output (noir-lang/noir#8635) fix: nargo expand reexports correctly implemented (noir-lang/noir#8693) feat(cli): Show multiple SSA passes (noir-lang/noir#8692) chore(test): Add regression test for defunctionalize (noir-lang/noir#8691) chore: add an assertion when parsing SSA that all functions are well-formed (noir-lang/noir#8671) feat!: prevent compiling blake3 hashes which barretenberg cannot prove (noir-lang/noir#8690) fix(ssa): Remove the array cache from the function inserter (noir-lang/noir#8607) chore: bump external pinned commits (noir-lang/noir#8684) chore: reenable fuzzing tests in CI (noir-lang/noir#8688) fix: Handle `&mut function` in defunctionalize (noir-lang/noir#8665) fix(defunctionalize): Higher order functions (HOF) dynamic dispatch and HOFs in arrays (noir-lang/noir#8672) chore(ci): avoid running fuzzer tests in the merge queue (noir-lang/noir#8664) fix: Make casts in `comptime` consistent with runtime casts (noir-lang/noir#8669) fix: relax connectedness requirement on purity analysis pass (noir-lang/noir#8667) fix: always use `u32` for indexing arrays in SSA (noir-lang/noir#8633) chore: explicitly pull from `next` branch in aztec-packages (noir-lang/noir#8660) chore(fuzz): Build the dictionary from the SSA (noir-lang/noir#8591) chore(docs): Remove old versioned docs (noir-lang/noir#8061) chore: bump external pinned commits (noir-lang/noir#8634) fix(SSA): don't use string literal if byte is "form feed" ('\f') (noir-lang/noir#8653) chore(defunctionalize): Add regression test for missing lambda variants panic (noir-lang/noir#8642) chore(ci): Do not run ast_fuzzer orig vs. morph in ci (noir-lang/noir#8646) fix: disable underflow fix for fields (noir-lang/noir#8631) fix: revert "fix: error on unused generic in trait impl (noir-lang/noir#8395)" (noir-lang/noir#8636) fix(ssa interpreter): Default to zero when we have an overflowing shl (noir-lang/noir#8638) fix: ensure that purity analysis pass explores all functions (noir-lang/noir#8452) feat: acir_formal_proofs (noir-lang/noir#8140) fix: error on unused generic in trait impl (noir-lang/noir#8395) fix(expand): use re-exports for non-visibile items (noir-lang/noir#8374) END_COMMIT_OVERRIDE --------- Co-authored-by: AztecBot <tech@aztecprotocol.com> Co-authored-by: Tom French <15848336+TomAFrench@users.noreply.github.com> Co-authored-by: Ary Borenszweig <asterite@gmail.com>

merge converging jmpifs for empty jmp chains

01cbe98

vezenovm requested a review from a team May 21, 2025 21:17

vezenovm added 3 commits May 21, 2025 17:17

Merge branch 'master' into mv/simplify-converging-jmpif

460ff8b

fmt

9bdf289

Merge remote-tracking branch 'origin/mv/simplify-converging-jmpif' in…

3084699

…to mv/simplify-converging-jmpif

TomAFrench changed the title ~~feat(simplify_cfg): Replace converging jmpif with empty block destinations with a single jmp~~ feat: Replace converging jmpif with empty block destinations with a single jmp May 22, 2025

michaeljklein approved these changes May 22, 2025

View reviewed changes

compiler/noirc_evaluator/src/ssa/opt/simplify_cfg.rs Outdated Show resolved Hide resolved

Update compiler/noirc_evaluator/src/ssa/opt/simplify_cfg.rs

ff10b51

Co-authored-by: Michael J Klein <michaeljklein@users.noreply.github.com>

github-actions bot reviewed May 28, 2025

View reviewed changes

vezenovm added 5 commits May 28, 2025 21:02

gate converging jmpif optimization to only be on Brillig

46ff845

Merge remote-tracking branch 'origin/mv/simplify-converging-jmpif' in…

0dca5a6

…to mv/simplify-converging-jmpif

Merge branch 'master' into mv/simplify-converging-jmpif

5609751

fix some tests with new validation

e277744

bump timeout

674c39a

vezenovm enabled auto-merge May 28, 2025 22:10

vezenovm added this pull request to the merge queue May 28, 2025

github-merge-queue bot pushed a commit that referenced this pull request May 28, 2025

feat: Replace converging jmpif with empty block destinations with a s…

ba81f5f

…ingle jmp (#8613) Co-authored-by: Michael J Klein <michaeljklein@users.noreply.github.com>

github-merge-queue bot removed this pull request from the merge queue due to failed status checks May 28, 2025

vezenovm added this pull request to the merge queue May 28, 2025

Merged via the queue into master with commit 00d636c May 28, 2025
118 checks passed

vezenovm deleted the mv/simplify-converging-jmpif branch May 28, 2025 23:26

noirwhal mentioned this pull request May 28, 2025

chore: Release Noir(1.0.0-beta.7) #8520

Merged

AztecBot mentioned this pull request May 29, 2025

chore: Bump Noir reference AztecProtocol/aztec-packages#14492

Merged

defkit pushed a commit that referenced this pull request May 29, 2025

feat: Replace converging jmpif with empty block destinations with a s…

8c6d14f

…ingle jmp (#8613) Co-authored-by: Michael J Klein <michaeljklein@users.noreply.github.com>

AztecBot mentioned this pull request Jun 1, 2025

chore(next): release 0.88.0 AztecProtocol/aztec-packages#14495

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: Replace converging jmpif with empty block destinations with a single jmp#8613

feat: Replace converging jmpif with empty block destinations with a single jmp#8613
vezenovm merged 10 commits intomasterfrom
mv/simplify-converging-jmpif

vezenovm commented May 21, 2025 •

edited by Savio-Sou

Loading

Uh oh!

github-actions bot commented May 21, 2025 •

edited

Loading

Uh oh!

github-actions bot commented May 21, 2025 •

edited

Loading

Uh oh!

TomAFrench commented May 21, 2025

Uh oh!

TomAFrench commented May 21, 2025 •

edited

Loading

Uh oh!

michaeljklein left a comment

Uh oh!

Uh oh!

github-actions bot left a comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

vezenovm commented May 21, 2025 • edited by Savio-Sou Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Problem*

Summary*

Additional Context

Documentation*

PR Checklist*

Uh oh!

github-actions bot commented May 21, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Changes to Brillig bytecode sizes

🧾 Summary (10% most significant diffs)

Uh oh!

github-actions bot commented May 21, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Changes to number of Brillig opcodes executed

🧾 Summary (10% most significant diffs)

Uh oh!

TomAFrench commented May 21, 2025

Uh oh!

TomAFrench commented May 21, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

michaeljklein left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

github-actions bot left a comment

Choose a reason for hiding this comment

⚠️ Performance Alert ⚠️

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

vezenovm commented May 21, 2025 •

edited by Savio-Sou

Loading

github-actions bot commented May 21, 2025 •

edited

Loading

github-actions bot commented May 21, 2025 •

edited

Loading

TomAFrench commented May 21, 2025 •

edited

Loading