Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

feat: refactor SSA passes to run on individual functions #6072

Merged
merged 16 commits into from
Oct 1, 2024
Merged

feat: refactor SSA passes to run on individual functions #6072

merged 16 commits into from
Oct 1, 2024

Conversation

michaeljklein
Copy link
Contributor

@michaeljklein michaeljklein commented Sep 17, 2024 •
edited
Loading

Description

Problem*

Resolves #5839

Refactors passes that only act on individual functions into impl Function { fn do_pass() }

Summary*

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.

michaeljklein and others added 12 commits September 17, 2024 17:16
@michaeljklein
wip: attempt SSA optimizations on zero-arg functions first, move per-…
4428074 
…function SSA optimizations into Function methods
@michaeljklein
Merge branch 'master' into michaeljklein/zero-arg-ssa-optimizations
76ee498
@michaeljklein
wip debugging non-idempotent passes, added pass for zero-arg functions
999231b
@michaeljklein
disable 0-arg constant folding for now because it's contradicting the…
ca540a3 
… inlining pass
@michaeljklein
cargo clippy/fmt
3476d4e
@michaeljklein
Merge branch 'master' into michaeljklein/zero-arg-ssa-optimizations
fcaef1b
@michaeljklein
disable array set optimizations when assumptions don't hold, cargo cl…
143358d 
…ippy/fmt
@michaeljklein
Merge branch 'master' into michaeljklein/zero-arg-ssa-optimizations
a7a6f18
@michaeljklein
only run 0-arg SSA pass on brillig functions
4621a59
@michaeljklein
re-enable reachable_blocks.len() == 1 assertion
5d9a4dc
@michaeljklein
Merge branch 'master' into michaeljklein/zero-arg-ssa-optimizations
80e59e5
@michaeljklein
Merge branch 'master' into michaeljklein/zero-arg-ssa-optimizations
24f076f
@michaeljklein michaeljklein changed the title feat: attempt SSA optimizations on zero-arg functions first feat: refactor SSA passes to run on individual functions Sep 30, 2024
@michaeljklein michaeljklein marked this pull request as ready for review October 1, 2024 14:02
@michaeljklein michaeljklein requested a review from a team October 1, 2024 14:02
@michaeljklein
re-remove bubble_up_constrains, add issue for combining Ssa::unroll_l…
4ccc9a1 
…oops_iteratively and Function::unroll_loops_iteratively
@TomAFrench TomAFrench added this pull request to the merge queue Oct 1, 2024
Merged via the queue into master with commit 85c502c Oct 1, 2024
47 checks passed
@TomAFrench TomAFrench deleted the michaeljklein/zero-arg-ssa-optimizations branch October 1, 2024 16:13
@noirwhal noirwhal mentioned this pull request Oct 1, 2024
Merged
AztecBot added a commit to AztecProtocol/aztec-packages that referenced this pull request Oct 2, 2024
@AztecBot
[1 changes] feat!: Syncing TypeVariableKind with Kind (noir-lang/noir…
886bbe1 
…#6094)

feat(perf): Simplify the cfg after DIE (noir-lang/noir#6184)
feat: refactor SSA passes to run on individual functions (noir-lang/noir#6072)
chore: Remove macros_api module (noir-lang/noir#6190)
fix: ensure to_bytes returns the canonical decomposition (noir-lang/noir#6084)
chore: rename `DefinitionKind::GenericType` (noir-lang/noir#6182)
@AztecBot AztecBot mentioned this pull request Oct 2, 2024
Open
TomAFrench added a commit that referenced this pull request Oct 2, 2024
@TomAFrench
Merge branch 'master' into aztec-packages
fc2b3f3 
* master: (36 commits)
  fix: ignore compression of blocks after msg.len in sha256_var (#6206)
  feat(perf): Optimize array set from get (#6207)
  chore(refactor): Array set optimization context struct for analysis (#6204)
  fix: type variables by default should have Any kind (#6203)
  feat: remove orphaned blocks from cfg to improve `simplify_cfg` pass. (#6198)
  fix(ssa): Check if result of array set is used in value of another array set (#6197)
  fix(docs): Rename recursion.md to recursion.mdx (#6195)
  feat: skip `remove_enable_side_effects` pass on brillig functions (#6199)
  feat!: Syncing TypeVariableKind with Kind (#6094)
  feat(perf): Simplify the cfg after DIE (#6184)
  feat: refactor SSA passes to run on individual functions (#6072)
  chore: Remove macros_api module (#6190)
  fix: ensure to_bytes returns the canonical decomposition (#6084)
  chore: rename `DefinitionKind::GenericType` (#6182)
  feat(perf): Remove redundant inc rc without instructions between (#6183)
  chore: reexport `CrateName` through `nargo` (#6177)
  feat(perf): Remove inc_rc instructions for arrays which are never mutably borrowed (#6168)
  chore(docs): Add link to more info about proving backend to Solidity verifier page (#5754)
  feat: let `Module::functions` and `Module::structs` return them in definition order (#6178)
  chore: split `test_program`s into modules (#6101)
  ...
github-merge-queue bot pushed a commit that referenced this pull request Oct 3, 2024
@noirwhal
chore: Release Noir(0.35.0) (#6030)
2a0d211 
🤖 I have created a release *beep* *boop*
---


<details><summary>0.35.0</summary>

## [0.35.0](v0.34.0...v0.35.0)
(2024-10-03)


### ⚠ BREAKING CHANGES

* Syncing TypeVariableKind with Kind
([#6094](#6094))
* remove sha256 opcode
(AztecProtocol/aztec-packages#4571)
* add support for u1 in the avm, ToRadix's radix arg is a memory addr
(AztecProtocol/aztec-packages#8570)
* Infer globals to be u32 when used in a type
([#6083](#6083))
* removing implicit numeric generics
([#5837](#5837))

### Features

* (LSP) if in runtime code, always suggest functions that return Quoted
as macro calls ([#6098](#6098))
([4a160cb](4a160cb))
* (LSP) remove unused imports
([#6129](#6129))
([98bc460](98bc460))
* (LSP) show global value on hover
([#6097](#6097))
([3d9d072](3d9d072))
* (LSP) suggest $vars inside `quote { ... }`
([#6114](#6114))
([73245b3](73245b3))
* Add `Expr::as_constructor`
([#5980](#5980))
([76dea7b](76dea7b))
* Add `Expr::as_for` and `Expr::as_for_range`
([#6039](#6039))
([abcae75](abcae75))
* Add `Expr::as_lambda`
([#6048](#6048))
([31130dc](31130dc))
* Add a `comptime` string type for string handling at compile-time
([#6026](#6026))
([5d2984f](5d2984f))
* Add support for u1 in the avm, ToRadix's radix arg is a memory addr
(AztecProtocol/aztec-packages#8570)
([e8bbce7](e8bbce7))
* Allow silencing an unused variable defined via `let`
([#6149](#6149))
([a2bc059](a2bc059))
* Allow visibility modifiers in struct definitions
([#6054](#6054))
([199be58](199be58))
* Check unconstrained trait impl method matches
([#6057](#6057))
([aedc983](aedc983))
* Default to outputting witness with file named after package
([#6031](#6031))
([e74b4ae](e74b4ae))
* Detect unconstructed structs
([#6061](#6061))
([bcb438b](bcb438b))
* Do not double error on import with error
([#6131](#6131))
([9b26650](9b26650))
* Expose `derived_generators` and `pedersen_commitment_with_separator`
from the stdlib ([#6154](#6154))
([877b806](877b806))
* Faster LSP by caching file managers
([#6047](#6047))
([c48a4f8](c48a4f8))
* Hoist constant allocation outside of loops
([#6158](#6158))
([180bfc9](180bfc9))
* Implement `to_be_radix` in the comptime interpreter
([#6043](#6043))
([1550278](1550278))
* Implement solver for mov_registers_to_registers
([#6089](#6089))
([4170c55](4170c55))
* Implement type paths
([#6093](#6093))
([2174ffb](2174ffb))
* Let `Module::functions` and `Module::structs` return them in
definition order
([#6178](#6178))
([dec9874](dec9874))
* Let LSP suggest macro calls too
([#6090](#6090))
([26d275b](26d275b))
* Let LSP suggest trait impl methods as you are typing them
([#6029](#6029))
([dfed81b](dfed81b))
* LSP autocompletion for `TypePath`
([#6117](#6117))
([3f79d8f](3f79d8f))
* **metaprogramming:** Add `#[use_callers_scope]`
([#6050](#6050))
([8c34046](8c34046))
* Optimize allocating immediate amounts of memory
(AztecProtocol/aztec-packages#8579)
([e8bbce7](e8bbce7))
* Optimize constraints in sha256
([#6145](#6145))
([164d29e](164d29e))
* **perf:** Allow array set last uses optimization in return block of
Brillig functions
([#6119](#6119))
([5598059](5598059))
* **perf:** Handle array set optimization across blocks for Brillig
functions ([#6153](#6153))
([12cb80a](12cb80a))
* **perf:** Optimize array set from get
([#6207](#6207))
([dfeb1c5](dfeb1c5))
* **perf:** Remove inc_rc instructions for arrays which are never
mutably borrowed
([#6168](#6168))
([a195442](a195442))
* **perf:** Remove redundant inc rc without instructions between
([#6183](#6183))
([be9dcfe](be9dcfe))
* **perf:** Remove unused loads in mem2reg and last stores per function
([#5925](#5925))
([19eef30](19eef30))
* **perf:** Remove useless paired RC instructions within a block during
DIE ([#6160](#6160))
([59c4118](59c4118))
* **perf:** Simplify the cfg after DIE
([#6184](#6184))
([a1b5046](a1b5046))
* Pretty print Quoted token stream
([#6111](#6111))
([cd81f85](cd81f85))
* Refactor SSA passes to run on individual functions
([#6072](#6072))
([85c502c](85c502c))
* Remove aztec macros
([#6087](#6087))
([9d96207](9d96207))
* Remove orphaned blocks from cfg to improve `simplify_cfg` pass.
([#6198](#6198))
([b4712c5](b4712c5))
* Remove sha256 opcode
(AztecProtocol/aztec-packages#4571)
([e8bbce7](e8bbce7))
* Remove unnecessary branching in keccak impl
([#6133](#6133))
([9c69dce](9c69dce))
* Represent assertions more similarly to function calls
([#6103](#6103))
([3ecd0e2](3ecd0e2))
* Show test output when running via LSP
([#6049](#6049))
([9fb010e](9fb010e))
* Simplify sha256 implementation
([#6142](#6142))
([acdfbbc](acdfbbc))
* Skip `remove_enable_side_effects` pass on brillig functions
([#6199](#6199))
([2303615](2303615))
* **ssa:** Simplify signed casts
([#6166](#6166))
([eec3a61](eec3a61))
* Swap endianness in-place in keccak implementation
([#6128](#6128))
([e3cdebe](e3cdebe))
* Syncing TypeVariableKind with Kind
([#6094](#6094))
([6440e18](6440e18))
* Visibility for globals
([#6161](#6161))
([103b54d](103b54d))
* Visibility for modules
([#6165](#6165))
([fcdbcb9](fcdbcb9))
* Visibility for traits
([#6056](#6056))
([5bbd9ba](5bbd9ba))
* Visibility for type aliases
([#6058](#6058))
([66d2a07](66d2a07))


### Bug Fixes

* (LSP) make goto and hover work well for attributes
([#6152](#6152))
([c679bc6](c679bc6))
* Allow macros to change types on each iteration of a comptime loop
([#6105](#6105))
([0864e7c](0864e7c))
* Allow providing default implementations of unconstrained trait methods
([#6138](#6138))
([7679bbc](7679bbc))
* Always parse all tokens from quoted token streams
([#6064](#6064))
([23ed74b](23ed74b))
* Be more lenient with semicolons on interned expressions
([#6062](#6062))
([052c4fe](052c4fe))
* Consider constants as used values to keep their rc ops
([#6122](#6122))
([1217005](1217005))
* Correct stack trace order in comptime assertion failures
([#6066](#6066))
([04f1636](04f1636))
* Databus panic for fns with empty params
(AztecProtocol/aztec-packages#8847)
([d252748](d252748))
* Decode databus return values
([#6095](#6095))
([c40eb1f](c40eb1f))
* Disable side-effects for no_predicates functions
([#6027](#6027))
([fc74c55](fc74c55))
* Disambiguate field or int static trait method call
([#6112](#6112))
([5b27ea4](5b27ea4))
* Do not duplicate constant arrays in brillig
([#6155](#6155))
([68f3022](68f3022))
* **docs:** Rename recursion.md to recursion.mdx
([#6195](#6195))
([054e48b](054e48b))
* Don't crash on untyped global used as array length
([#6076](#6076))
([426f295](426f295))
* Ensure to_bytes returns the canonical decomposition
([#6084](#6084))
([b280a79](b280a79))
* Error on `&mut x` when `x` is not mutable
([#6037](#6037))
([57afc7d](57afc7d))
* Fix canonicalization bug
([#6033](#6033))
([7397772](7397772))
* Fix comptime type formatting
([#6079](#6079))
([e678091](e678091))
* Handle multi-byte utf8 characters in formatter
([#6118](#6118))
([b1d0619](b1d0619))
* Handle parenthesized expressions in array length
([#6132](#6132))
([9f0b397](9f0b397))
* Ignore compression of blocks after msg.len in sha256_var
([#6206](#6206))
([76eec71](76eec71))
* Infer globals to be u32 when used in a type
([#6083](#6083))
([78262c9](78262c9))
* Initialise databus using return values
([#6074](#6074))
([e17dfa5](e17dfa5))
* Let LSP suggest fields and methods in LValue chains
([#6051](#6051))
([5bf6567](5bf6567))
* Let token pretty printer handle `+=` and similar token sequences
([#6135](#6135))
([684b6cc](684b6cc))
* **mem2reg:** Remove possibility of underflow
([#6107](#6107))
([aea5cc7](aea5cc7))
* Parse a statement as an expression
([#6040](#6040))
([ab203e4](ab203e4))
* Pass radix directly to the blackbox
([#6164](#6164))
([82b89c4](82b89c4))
* Preserve generic kind on trait methods
([#6099](#6099))
([1df102a](1df102a))
* Prevent check_can_mutate crashing on undefined variable
([#6044](#6044))
([b3accfc](b3accfc))
* Revert mistaken stack size change
([#6212](#6212))
([a37117a](a37117a))
* **ssa:** Check if result of array set is used in value of another
array set ([#6197](#6197))
([594ec91](594ec91))
* **ssa:** RC correctness issue
([#6134](#6134))
([5b1c896](5b1c896))
* Type variables by default should have Any kind
([#6203](#6203))
([268f2a0](268f2a0))
* Unify macro result type with actual type
([#6086](#6086))
([af52873](af52873))
* Update databus in flattening
([#6063](#6063))
([e993da1](e993da1))


### Miscellaneous Chores

* Removing implicit numeric generics
([#5837](#5837))
([eda9043](eda9043))
</details>

<details><summary>0.51.0</summary>

## [0.51.0](v0.50.0...v0.51.0)
(2024-10-03)


### ⚠ BREAKING CHANGES

* remove sha256 opcode
(AztecProtocol/aztec-packages#4571)
* add support for u1 in the avm, ToRadix's radix arg is a memory addr
(AztecProtocol/aztec-packages#8570)
* Add Not instruction in brillig
(AztecProtocol/aztec-packages#8488)
* **avm:** variants for SET opcode
(AztecProtocol/aztec-packages#8441)
* **avm/brillig:** take addresses in calldatacopy
(AztecProtocol/aztec-packages#8388)
* constant inputs for blackbox
(AztecProtocol/aztec-packages#7222)

### Features

* (bb) 128-bit challenges
(AztecProtocol/aztec-packages#8406)
([3c3ed1e](3c3ed1e))
* **acir_gen:** Width aware ACIR gen addition
([#5493](#5493))
([85fa592](85fa592))
* Add assertions for ACVM `FunctionInput` `bit_size`
([#5864](#5864))
([8712f4c](8712f4c))
* Add Not instruction in brillig
(AztecProtocol/aztec-packages#8488)
([95e19ab](95e19ab))
* Add recursive aggregation object to proving/verification keys
(AztecProtocol/aztec-packages#6770)
([4ea25db](4ea25db))
* Add reusable procedures to brillig generation
(AztecProtocol/aztec-packages#7981)
([5c4f19f](5c4f19f))
* Add support for u1 in the avm, ToRadix's radix arg is a memory addr
(AztecProtocol/aztec-packages#8570)
([e8bbce7](e8bbce7))
* Added indirect const instruction
(AztecProtocol/aztec-packages#8065)
([5c4f19f](5c4f19f))
* Adding aggregation to honk and rollup
(AztecProtocol/aztec-packages#7466)
([4ea25db](4ea25db))
* Automate verify_honk_proof input generation
(AztecProtocol/aztec-packages#8092)
([5c4f19f](5c4f19f))
* **avm/brillig:** Take addresses in calldatacopy
(AztecProtocol/aztec-packages#8388)
([3c3ed1e](3c3ed1e))
* **avm:** Variants for SET opcode
(AztecProtocol/aztec-packages#8441)
([3c3ed1e](3c3ed1e))
* Avoid heap allocs when going to/from field
(AztecProtocol/aztec-packages#7547)
([daad75c](daad75c))
* Change the layout of arrays and vectors to be a single pointer
(AztecProtocol/aztec-packages#8448)
([d4832ec](d4832ec))
* Constant inputs for blackbox
(AztecProtocol/aztec-packages#7222)
([fb97bb9](fb97bb9))
* Hook up secondary calldata column in dsl
(AztecProtocol/aztec-packages#7759)
([4ea25db](4ea25db))
* Integrate new proving systems in e2e
(AztecProtocol/aztec-packages#6971)
([daad75c](daad75c))
* Make Brillig do integer arithmetic operations using u128 instead of
Bigint (AztecProtocol/aztec-packages#7518)
([daad75c](daad75c))
* Make token transfer be recursive
(AztecProtocol/aztec-packages#7730)
([4ea25db](4ea25db))
* New test programs for wasm benchmarking
(AztecProtocol/aztec-packages#8389)
([95e19ab](95e19ab))
* Note hashes as points
(AztecProtocol/aztec-packages#7618)
([4ea25db](4ea25db))
* Optimize allocating immediate amounts of memory
(AztecProtocol/aztec-packages#8579)
([e8bbce7](e8bbce7))
* Optimize constant array handling in brillig_gen
(AztecProtocol/aztec-packages#7661)
([4ea25db](4ea25db))
* Optimize to_radix
(AztecProtocol/aztec-packages#8073)
([5c4f19f](5c4f19f))
* Pass calldata ids to the backend
(AztecProtocol/aztec-packages#7875)
([4ea25db](4ea25db))
* Poseidon2 gates for Ultra arithmetisation
(AztecProtocol/aztec-packages#7494)
([5c4f19f](5c4f19f))
* **profiler:** Add support for brillig functions in opcodes-flamegraph
(AztecProtocol/aztec-packages#7698)
([4ea25db](4ea25db))
* Remove sha256 opcode
(AztecProtocol/aztec-packages#4571)
([e8bbce7](e8bbce7))
* Removing superfluous call to MSM
(AztecProtocol/aztec-packages#7708)
([4ea25db](4ea25db))
* Report gates and VKs of private protocol circuits with megahonk
(AztecProtocol/aztec-packages#7722)
([4ea25db](4ea25db))
* Simplify constant calls to `poseidon2_permutation`, `schnorr_verify`
and `embedded_curve_add`
([#5140](#5140))
([2823ba7](2823ba7))
* Small optimization in toradix
(AztecProtocol/aztec-packages#8040)
([5c4f19f](5c4f19f))
* Sync from noir
(AztecProtocol/aztec-packages#7392)
([fb97bb9](fb97bb9))
* Sync from noir
(AztecProtocol/aztec-packages#7400)
([fb97bb9](fb97bb9))
* Sync from noir
(AztecProtocol/aztec-packages#7432)
([daad75c](daad75c))
* Sync from noir
(AztecProtocol/aztec-packages#7444)
([daad75c](daad75c))
* Sync from noir
(AztecProtocol/aztec-packages#7454)
([daad75c](daad75c))
* Sync from noir
(AztecProtocol/aztec-packages#7512)
([daad75c](daad75c))
* Sync from noir
(AztecProtocol/aztec-packages#7577)
([daad75c](daad75c))
* Sync from noir
(AztecProtocol/aztec-packages#7583)
([daad75c](daad75c))
* Sync from noir
(AztecProtocol/aztec-packages#7743)
([4ea25db](4ea25db))
* Sync from noir
(AztecProtocol/aztec-packages#7862)
([4ea25db](4ea25db))
* Sync from noir
(AztecProtocol/aztec-packages#7945)
([4ea25db](4ea25db))
* Sync from noir
(AztecProtocol/aztec-packages#7958)
([5c4f19f](5c4f19f))
* Sync from noir
(AztecProtocol/aztec-packages#8008)
([5c4f19f](5c4f19f))
* Sync from noir
(AztecProtocol/aztec-packages#8093)
([5c4f19f](5c4f19f))
* Sync from noir
(AztecProtocol/aztec-packages#8125)
([f0c2686](f0c2686))
* Sync from noir
(AztecProtocol/aztec-packages#8237)
([f0c2686](f0c2686))
* Sync from noir
(AztecProtocol/aztec-packages#8423)
([3c3ed1e](3c3ed1e))
* Sync from noir
(AztecProtocol/aztec-packages#8435)
([3c3ed1e](3c3ed1e))
* Sync from noir
(AztecProtocol/aztec-packages#8466)
([3c3ed1e](3c3ed1e))
* Sync from noir
(AztecProtocol/aztec-packages#8482)
([d4832ec](d4832ec))
* Sync from noir
(AztecProtocol/aztec-packages#8512)
([95e19ab](95e19ab))
* Sync from noir
(AztecProtocol/aztec-packages#8526)
([95e19ab](95e19ab))
* TXE nr deployments, dependency cleanup for CLI
(AztecProtocol/aztec-packages#7548)
([4ea25db](4ea25db))
* Typing return values of embedded_curve_ops
(AztecProtocol/aztec-packages#7413)
([daad75c](daad75c))
* Unify all acir recursion constraints based on RecursionConstraint and
proof_type (AztecProtocol/aztec-packages#7993)
([5c4f19f](5c4f19f))


### Bug Fixes

* Add trailing extra arguments for backend in gates_flamegraph
(AztecProtocol/aztec-packages#7472)
([daad75c](daad75c))
* **debugger:** Update the debugger to handle the new Brillig debug
metadata format ([#5706](#5706))
([a31f82e](a31f82e))
* Deflatten databus visibilities
(AztecProtocol/aztec-packages#7761)
([4ea25db](4ea25db))
* Do not duplicate redundant Brillig debug metadata
([#5696](#5696))
([e4f7dbe](e4f7dbe))
* Export brillig names in contract functions
(AztecProtocol/aztec-packages#8212)
([f0c2686](f0c2686))
* Handle multiple entry points for Brillig call stack resolution after
metadata deduplication
([#5788](#5788))
([38fe9dd](38fe9dd))
* Move BigInt modulus checks to runtime in brillig
([#5374](#5374))
([741d339](741d339))
* Restrict keccak256_injective test input to 8 bits
([#5977](#5977))
([a1b1346](a1b1346))
* Revert "feat: Sync from noir
(AztecProtocol/aztec-packages#7512)"
(AztecProtocol/aztec-packages#7558)
([daad75c](daad75c))
* Runtime brillig bigint id assignment
([#5369](#5369))
([a8928dd](a8928dd))
</details>

---
This PR was generated with [Release
Please](https://github.com/googleapis/release-please). See
[documentation](https://github.com/googleapis/release-please#release-please).
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@TomAFrench TomAFrench TomAFrench approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

Attempt SSA optimizations on zero-argument functions first
2 participants
@michaeljklein @TomAFrench