feat: no longer recompute vk's in CIVC proofs #13590

ludamad · 2025-04-16T00:00:55Z

Apologies for the big PR. Lots of the stuff in here ended up being chicken and egg with wanting to do refactoring to make this process smoother/help debug tricky issues, and wanting to see that those refactorings actually make sense by the end.

We no longer compute VKs on the fly in CIVC. This saves ~25% of computation. This is done throughout by consolidating IVC inputs into a single ivc-inputs.msgpack structure which supports passing bytecode, witness & vk information to the bb backend. Now attaches a name for each function, as well.

Major features:

IVC inputs passed thru native and wasm are always passed a single file/buffer. This is encoded using msgpack and capture bytecode, witness, vk, and function name (which is now printed, but only properly passed by native) For native, the bincode and witnesses are gzipped, for WASM they are uncompressed. For actions such as gates or write_vk, the IVC inputs are used with same structure but witness and vk data can be blank.

This has a bunch of implications, such as having to break away from the rigid API base class in bb cli (which overall doesn't feel worthwhile anyway as CIVC is fundamentally different than UH), having to string vk info along, etc.

Other features:

Debuggability:

Correct README.md instructions on WASM stack traces (give up on getting line numbers working :/)
clangd now properly shows all errors in a C++ file you're browsing, instead of only showing you the first error.

Cleanup

small cleanup to acir tests, but still not testing new ivc flow there. Lightest weight test is ivc-integration in yarn-project
Get rid of --input_type in bb cli for CIVC. now implied always to be what was previously runtime_stack. Simplifies usages, other modes were unused.
more ignored linting in the clangd file. Maybe one day we can enforce the remaining as errors.
Clean up msgpack usage. Msgpack headers were leaking everywhere and it is a chunky library.
Consolidate with using msgpackr as our only typescript messagepack library

Benches

use wasmtime helper in bb bootstrap. deduplicate code in bench. bench now honours NATIVE_PRESET, and if you do

export NATIVE_PRESET=op-count-time
./bootstrap.sh
./bootstrap.sh bench

you will get op count timings for our native ivc benches.

…fier-keys

…akdown

ludamad · 2025-04-17T14:47:58Z

barretenberg/exports.json

@@ -128,23 +128,23 @@
    "isAsync": false
  },
  {
-    "functionName": "blake2s",


hmm not sure why this file isnt git-ignored tbh

probably out of scope of this already overscoped PR tho

ludamad · 2025-04-17T14:48:54Z

barretenberg/ts/package.json

@@ -89,6 +88,7 @@
    "html-webpack-plugin": "^5.5.1",
    "idb-keyval": "^6.2.1",
    "jest": "^29.5.0",
+    "msgpackr": "^1.11.2",


consolidation of msgpack libs used

ludamad · 2025-04-17T14:49:39Z

barretenberg/ts/src/barretenberg/backend.ts

+      functionName,
+    });
+  }
+  return new Encoder({ useRecords: false }).pack(steps);


useRecords: false just means 'dont do fancy key name compression' - needed by our C++ msgpack lib

ludamad · 2025-04-17T14:50:35Z

barretenberg/ts/src/barretenberg/schnorr.test.ts

@@ -1,182 +0,0 @@
-import { TextEncoder } from 'util';
-import { Buffer128, Buffer32, Fq, Fr, Point } from '../types/index.js';
-import { Barretenberg } from './index.js';


No longer exposed via bindings. I had trouble generating auto-bindings for it and it is unused logic

ludamad · 2025-04-17T14:51:32Z

barretenberg/ts/src/barretenberg_api/index.ts

@@ -12,26 +12,6 @@ import {
  OutputType,
 } from '../serialize/index.js';
 import { Fr, Fq, Point, Buffer32, Buffer128, Ptr } from '../types/index.js';


everyone was editing the file that said DO NOT EDIT. I have finally regenerated it. Luckily mostly people kept to the pattern

ludamad · 2025-04-17T14:51:54Z

barretenberg/ts/src/bindgen/mappings.ts

@@ -13,6 +13,7 @@ const typeMap: { [key: string]: string } = {
  'fq::vec_in_buf': 'Fq[]',
  'fq::vec_out_buf': 'Fq[]',
  'const uint8_t *': 'Uint8Array',
+  'uint8_vec_vec_in_buf': 'Uint8Array[]',


ended up unused, but meh

ludamad · 2025-04-17T14:52:14Z

barretenberg/ts/src/bindgen/typescript.ts

@@ -9,6 +9,7 @@ export function generateTypeScriptCode(filename: string) {

  let output = `// WARNING: FILE CODE GENERATED BY BINDGEN UTILITY. DO NOT EDIT!
 /* eslint-disable @typescript-eslint/no-unused-vars */
+import { BarretenbergWasmMain } from '../barretenberg_wasm/barretenberg_wasm_main/index.js';
 import { BarretenbergWasmWorker, BarretenbergWasm } from '../barretenberg_wasm/index.js';
 import { BufferDeserializer, NumberDeserializer, VectorDeserializer, BoolDeserializer, StringDeserializer, serializeBufferable, OutputType } from '../serialize/index.js';


Capture a manual edit that had been made

ludamad · 2025-04-17T14:52:49Z

barretenberg/ts/src/main.ts

-    const verified = await api.acirProveAndVerifyAztecClient(bytecode, witness);
-    debug(`Verification ${verified ? 'successful' : 'failed'}`);
-    return verified;
-  } finally {


Just removing CIVC from the CLI as it would have been maintenance effort

ludamad · 2025-04-17T14:53:14Z

barretenberg/ts/tsconfig.json

@@ -12,6 +12,7 @@
    "declarationMap": true,
    "importHelpers": true,
    "resolveJsonModule": true,


msgpackr didnt type check, but doesnt affect much. this makes compilation faster, too

ludamad · 2025-04-17T14:54:01Z

yarn-project/bb-prover/scripts/generate_civc_vks.sh


-pinned_civc_inputs_url="https://aztec-ci-artifacts.s3.us-east-2.amazonaws.com/protocol/bb-civc-inputs-v2.tar.gz"
+pinned_civc_inputs_url="https://aztec-ci-artifacts.s3.us-east-2.amazonaws.com/protocol/bb-civc-inputs-v3.tar.gz"


bump as this stuff is now formatted as a single file ivc-inputs.msgpack

ludamad · 2025-04-17T14:55:26Z

yarn-project/scripts/run_test.sh

@@ -30,11 +31,11 @@ if [ "${ISOLATE:-0}" -eq 1 ]; then
    -e NODE_OPTIONS="--no-warnings --experimental-vm-modules --loader @swc-node/register" \
    -e LOG_LEVEL \


allows filtering the exact test by passing args along

ledwards2225

Overall looks great. Lots of good cleanup in here. As always, I think its worth striving for more incremental PRs but I can see how the scope of this one could have naturally evolved as you went.

It seems like this raises questions about the CIVC API, the extent to which it should adhere to something modeled after UH or whether it should exist at all.

ledwards2225 · 2025-04-17T14:26:52Z

barretenberg/acir_tests/bootstrap.sh

@@ -215,9 +215,6 @@ function run_benchmark {

 # TODO(https://github.com/AztecProtocol/barretenberg/issues/1254): More complete testing, including failure tests
 function bench {
-  # TODO(https://github.com/AztecProtocol/barretenberg/issues/1265) fix acir benchmarking


Should this issue be closed? I see it doesn't have a description

Done. it was the old acir benchmarks. they're probably not coming back at this point

ledwards2225 · 2025-04-17T14:30:24Z

barretenberg/cpp/bootstrap.sh

@@ -300,7 +298,7 @@ case "$cmd" in
    # Takes an optional master commit to download them from. Otherwise, downloads from latest master commit.
    git fetch origin master

-    build the benchmarked benches
+    # build the benchmarked benches


huh.. how was this working

it calls 'build' simply :) its valid BASH

ledwards2225 · 2025-04-17T14:33:05Z

barretenberg/cpp/src/barretenberg/api/api.hpp

@@ -53,21 +51,11 @@ class API {
                       const std::filesystem::path& bytecode_path,
                       const std::filesystem::path& witness_path) = 0;

-    virtual void prove(const Flags& flags,


Flag - This is surprising

Ok I see, just not defining it as part of the interface anymore. I suppose once we get rid of plonk there's very little value to defining this interface at all if UH and CIVC dont both adhere to something

plonk doesnt even adhere to this

ledwards2225 · 2025-04-17T14:39:37Z

barretenberg/cpp/src/barretenberg/bb/cli.cpp

@@ -237,6 +227,12 @@ int parse_and_run_cli_command(int argc, char* argv[])
            /* ->check(CLI::ExistingFile) OR stdin indicator - */;
    };

+    const auto add_ivc_inputs_path_options = [&](CLI::App* subcommand) {


Not directly related to this change but this does make me think: what exactly is the value in having a CLI for CIVC? To the extent there is one, it seems to be for internal testing only.

it's valuable right now for power users who can use aztec.js with a native bb and get speedups. It's probably not something we should get rid of until we have another native way to expose this functionality

ledwards2225 · 2025-04-17T14:42:33Z

barretenberg/cpp/src/barretenberg/bb/cli.cpp

@@ -698,8 +694,9 @@ int parse_and_run_cli_command(int argc, char* argv[])
        print_subcommand_options(deepest);
    }

-    // prob this construction is too much
-    const auto execute_command = [&](API& api) {
+    // TODO(AD): it is inflexible that CIVC shares an API command (prove) with UH this way. The base API class is a


I think Cody had an overly optimistic view of the extent to which it made sense to uniformize across "proving systems" here. At the end of the day it will just be UH and CIVC and as I alluded to, I'm not sure I see any real benefit to a CIVC CLI. Much less one that adheres to the interface of UH

ledwards2225 · 2025-04-17T14:46:32Z

barretenberg/cpp/src/barretenberg/client_ivc/private_execution_steps.cpp

+            acir_format::circuit_buf_to_acir_format(bytecode_buf, /*honk_recursion=*/0);
+        acir_format::WitnessVector witness = acir_format::witness_buf_to_witness_data(witness_buf);
+
+        folding_stack.push_back({ constraints, witness });


I don't think you introduced this pattern but this is probably an amount of copying that's worth avoiding. These things can be pretty large

I posted about this in noir-lang trying to resolve the root cause - bincode expecting fresh uint8_t vectors

ledwards2225 · 2025-04-17T14:53:16Z

yarn-project/bb-prover/src/wasm/bb_wasm_private_kernel_prover.ts

@@ -42,6 +45,7 @@ export abstract class BBWASMPrivateKernelProver extends BBPrivateKernelProver {
  }

  public override async computeGateCountForCircuit(_bytecode: Buffer, _circuitName: string): Promise<number> {
+    // Note we do not pass the vk to the backend. This is unneeded for gate counts.


Eventually we should just store gate counts along with the VK and such a command will not be needed

🤖 I have created a new Aztec Packages release --- ## [0.86.0](v0.85.0...v0.86.0) (2025-04-24) ### ⚠ BREAKING CHANGES * Use combined p2p and http prover coordination ([#13760](#13760)) * use single extended viem client ([#13715](#13715)) * remove slice read from CALL ([#13729](#13729)) ### Features * `#[allow(dead_code)]` (noir-lang/noir#8066) ([366d980](366d980)) * add aztec-up amm test w/ proving & sponsoredfpc ([#13209](#13209)) ([f54c8a1](f54c8a1)) * Add Rollup to registry and produce blocks ([#13556](#13556)) ([9ed10d3](9ed10d3)) * adding devnet tests ([#12808](#12808)) ([7687a1f](7687a1f)) * audit tracking tooling ([#13639](#13639)) ([e50d8e0](e50d8e0)) * **avm:** fast entity indexing without macros ([#13737](#13737)) ([5c8a993](5c8a993)) * **avm:** quick n dirty memory trace ([#13659](#13659)) ([062c6a9](062c6a9)) * **avm:** support col arrays and add keccak cols ([#13711](#13711)) ([9c4345a](9c4345a)) * **avm:** tagged value type in C++ ([#13540](#13540)) ([b30b5b3](b30b5b3)) * avoid unnecessary zero check in brillig overflow check (noir-lang/noir#8109) ([366d980](366d980)) * **contracts:** static + snapshotted validator set ([#13046](#13046)) ([90033f2](90033f2)) * enable mempool limiter on devnet ([#13722](#13722)) ([36fb0a1](36fb0a1)) * exec opcode spec table ([#13594](#13594)) ([d4d503a](d4d503a)) * **experimental:** Implement separate `-Zownership` analysis for ownership pass (noir-lang/noir#7861) ([366d980](366d980)) * expose mempool limit through helm chart ([#13616](#13616)) ([822e8d9](822e8d9)) * Garaga UltraStarknet[Zk]Honk flavours ([#11489](#11489)) ([6bc34a1](6bc34a1)) * IPA claim as public component ([#13429](#13429)) ([1ce3c43](1ce3c43)) * no longer recompute vk's in CIVC proofs ([#13590](#13590)) ([66a61ba](66a61ba)) * **p2p:** add private peers ([#12585](#12585)) ([4264c8c](4264c8c)) * **p2p:** optional P2P_BROADCAST_PORT ([#13525](#13525)) ([f709fab](f709fab)) * Port callstack resolution from aztec to noirjs (noir-lang/noir#7642) ([366d980](366d980)) * Refactor IPA claim handling in acir format to support them for AVM ([#13547](#13547)) ([df73c05](df73c05)) * remove slice read from CALL ([#13729](#13729)) ([7e7eb85](7e7eb85)) * replace field divisions by constants with multiplication by inv… (noir-lang/noir#8053) ([366d980](366d980)) * report world state size on disk ([#13706](#13706)) ([0dd8a7e](0dd8a7e)) * skip inverting field elements with values +-1 in ACVM (noir-lang/noir#8049) ([44c1347](44c1347)) * SMT Verificaiton Module: Data Structures ([#13658](#13658)) ([2de3bc5](2de3bc5)) * SMT verification module updates ([#13551](#13551)) ([f02123d](f02123d)) * ssa fuzzer (noir-lang/noir#7641) ([366d980](366d980)) * Tighter timing on ACVM ([#13743](#13743)) ([45cd39b](45cd39b)) * track rewards and slots ([#13546](#13546)) ([489c6cf](489c6cf)) * Use combined p2p and http prover coordination ([#13760](#13760)) ([4d04e62](4d04e62)) * VK generation test for HonkRecursionConstraint ([#13637](#13637)) ([71e81ce](71e81ce)) ### Bug Fixes * **acir:** Check whether opcodes were laid down for non-equality check before fetching payload locations (noir-lang/noir#8133) ([366d980](366d980)) * add proper handling for `u128` in comptime interpreter (noir-lang/noir#8079) ([366d980](366d980)) * Add version to gossipsub protocol ([#13567](#13567)) ([8ce4829](8ce4829)) * allow keywords in function names in SSA parser (noir-lang/noir#8063) ([44c1347](44c1347)) * Allow more slack in AST calibration for CI (noir-lang/noir#8076) ([366d980](366d980)) * amm bot ([#13553](#13553)) ([6b93832](6b93832)), closes [#13544](#13544) * **avm:** cpp addressing ([#13652](#13652)) ([9d941b6](9d941b6)) * blob sink waits for reg address when needed ([#13628](#13628)) ([486a42a](486a42a)) * **brillig:** SliceRefCount reads from the appropriate pointer (noir-lang/noir#8148) ([366d980](366d980)) * Check against multiple aws profiles in build instance credentails ([#13421](#13421)) ([71a408a](71a408a)) * dependency cycles in public simulator - part 0 (sim -> context) ([#13678](#13678)) ([e00089f](e00089f)) * dependency cycles in public simulator - part 1 (errors/revertReason) ([#13679](#13679)) ([c75e365](c75e365)) * dependency cycles in public simulator part 2 (serializable bytecode) ([#13680](#13680)) ([5d3e24c](5d3e24c)) * discv5 test failure ([#13653](#13653)) ([58c143b](58c143b)) * **docs:** Make viewing (outdated) protocol specs work locally ([#13534](#13534)) ([694537d](694537d)) * Increase timeout for p2p integration test ([#13720](#13720)) ([c244b2e](c244b2e)) * mac bb --version ([#13690](#13690)) ([9f8b64f](9f8b64f)) * make translator use ultra rather than eccvm ops ([#13489](#13489)) ([53c070d](53c070d)) * more worker fixes ([#13759](#13759)) ([080485a](080485a)) * move helm lint to build ([#13532](#13532)) ([7768ff8](7768ff8)) * native ivc benches not publishing ([#13665](#13665)) ([7ab712d](7ab712d)) * no exports of simulator should depend on jest-mock-extended ([#13694](#13694)) ([65a9f33](65a9f33)) * only clear trailing zeros on logs ([#13727](#13727)) ([e392d7c](e392d7c)) * **p2p:** better batch connection sampling ([#13674](#13674)) ([9c1d9f1](9c1d9f1)) * **p2p:** reqresp types + batch request tx pool filtering ([#13666](#13666)) ([8e2a3c9](8e2a3c9)) * **parser:** error on missing let semicolon in trait (and others) (noir-lang/noir#8101) ([366d980](366d980)) * pass along coinbase ([#13560](#13560)) ([593f810](593f810)), closes [#13643](#13643) * prover-node start ([#13627](#13627)) ([5d87f7a](5d87f7a)) * quote numbers ([#13654](#13654)) ([f0d1ec1](f0d1ec1)) * remove all txs from a failed epoch ([#13771](#13771)) ([abc4623](abc4623)) * remove insecure dummy round derivation from sumcheck and shplemini ([#13488](#13488)) ([9a3bb46](9a3bb46)) * replace values in data_bus too (noir-lang/noir#8086) ([366d980](366d980)) * retry deploy npm ([#13691](#13691)) ([76cab3f](76cab3f)) * run yarn ([#13713](#13713)) ([0e51259](0e51259)) * squash nightly foundry warning ([#13640](#13640)) ([e21f9fd](e21f9fd)) * SSA pass print filter to include the count (noir-lang/noir#8074) ([366d980](366d980)) * **ssa:** Do not inline simple recursive functions (noir-lang/noir#8127) ([366d980](366d980)) * **ssa:** Do not unroll loop with break (noir-lang/noir#8090) ([366d980](366d980)) * **ssa:** Do not use inner else condition when collapsing IfElse based upon the then condition (noir-lang/noir#8040) ([44c1347](44c1347)) * **ssa:** fix possibility to `Field % Field` operaions in Brillig from SSA (noir-lang/noir#8105) ([366d980](366d980)) * **ssa:** Loop range with u1 (noir-lang/noir#8131) ([366d980](366d980)) * update metric name to avoid conflicts ([#13629](#13629)) ([70c58ab](70c58ab)), closes [#13626](#13626) * wasm memory benchmark ([#13573](#13573)) ([9a73c4a](9a73c4a)) * wrapping mul support for u128 (noir-lang/noir#7941) ([366d980](366d980)) ### Miscellaneous * Accept optional test path for emitting integration tests (noir-lang/noir#8062) ([366d980](366d980)) * Add `GITHUB_TOKEN` to more workflow steps (noir-lang/noir#8046) ([44c1347](44c1347)) * add a benchmark for opcodes which need a batchable inversion (noir-lang/noir#8110) ([366d980](366d980)) * add hacky faster bootstrap for bb-centric e2e flow ([#13587](#13587)) ([498d433](498d433)) * address some visibility warnings ([#13728](#13728)) ([35dec90](35dec90)) * **avm:** less codegen for lookups ([#13741](#13741)) ([ec468d9](ec468d9)) * **avm:** move interaction jobs to trace builders ([#13621](#13621)) ([caac1c9](caac1c9)) * bb.js non-inlined web workers ([#13736](#13736)) ([0e604a1](0e604a1)) * **bb:** debugging helpers ([#13584](#13584)) ([0ebb29e](0ebb29e)) * **bb:** Make goblin a proper source module ([#13580](#13580)) ([3048a14](3048a14)) * bespoke export for client native prover / PXE server store lazy load ([#13783](#13783)) ([640dd08](640dd08)) * Blob and archiver syncing improvements ([#13542](#13542)) ([629aa2a](629aa2a)), closes [#13530](#13530) * blob archive client factory code ([#13648](#13648)) ([7b16791](7b16791)) * bug list ([#13773](#13773)) ([c6d4fb8](c6d4fb8)) * bump axois ([#13453](#13453)) ([1c20920](1c20920)) * bump bb to 0.84.0 (noir-lang/noir#8047) ([44c1347](44c1347)) * bump bignum timeout (noir-lang/noir#8080) ([366d980](366d980)) * bump external pinned commits (noir-lang/noir#8045) ([44c1347](44c1347)) * **ci:** run non dependent bootstrap build steps in parallel ([#13454](#13454)) ([b17e18d](b17e18d)) * cleanup aztec up amm test ([#13754](#13754)) ([3d92afe](3d92afe)) * Cleanup scripts in package jsons in yarn-project ([#13527](#13527)) ([a296945](a296945)) * **contracts:** core/staking -> core/slashing ([#13748](#13748)) ([5d93a0e](5d93a0e)) * create module for array handling in acirgen (noir-lang/noir#8119) ([366d980](366d980)) * dashboards use ${data_source} variable ([#13613](#13613)) ([45533d9](45533d9)) * delete honk_recursion for building ACIR ([#13664](#13664)) ([ac95729](ac95729)) * delete Ultra Vanilla CIVC ([#13669](#13669)) ([ae578a2](ae578a2)) * delete zeromorph ([#13667](#13667)) ([c8acae0](c8acae0)) * Disable blobscan by default ([#13742](#13742)) ([c55088e](c55088e)) * **docs:** `aztec` and `aztec-wallet` cli reference updates ([#13692](#13692)) ([d11d154](d11d154)) * **docs:** add 0.85.0 docs ([#13578](#13578)) ([cf07358](cf07358)) * **docs:** add AND/XOR bug to bug list ([#13784](#13784)) ([797ae6d](797ae6d)) * **docs:** Add architecture docs (noir-lang/noir#7992) ([366d980](366d980)) * **docs:** adding redirects for node guide ([#13768](#13768)) ([e96e5bf](e96e5bf)) * **docs:** bugfix + deploy on master ([#13717](#13717)) ([010975a](010975a)) * **docs:** Fix some stuff on fees, rm confusing storage example ([#13501](#13501)) ([c00b20b](c00b20b)) * **docs:** Sequencer and Prover Guide ([#13433](#13433)) ([55b9a2b](55b9a2b)) * **docs:** update bb commands to match 0.84.0 (noir-lang/noir#8050) ([366d980](366d980)) * **docs:** Update docs bootrap to to include release ([#13615](#13615)) ([2154d5e](2154d5e)) * don't use `set_from_value_id` in `remove_if_else` (noir-lang/noir#8070) ([366d980](366d980)) * don't use `set_value_from_id` in `as_slice_length` (noir-lang/noir#8039) ([44c1347](44c1347)) * don't use `set_value_from_id` in `constant_folding` (noir-lang/noir#8091) ([366d980](366d980)) * don't use `set_value_from_id` in `loop_invariant` (noir-lang/noir#8085) ([366d980](366d980)) * don't use `set_value_from_id` in `remove_bit_shifts` (noir-lang/noir#8071) ([366d980](366d980)) * don't use `set_value_from_id` in `simplify_cfg` (noir-lang/noir#8072) ([366d980](366d980)) * **experimental:** Function::simple_optimization for SSA optimizations (noir-lang/noir#8102) ([366d980](366d980)) * Fetch rollup address using version as index ([#13620](#13620)) ([60e73f9](60e73f9)) * Fix license files ([#13695](#13695)) ([9931fb8](9931fb8)) * mempool limit ([#13735](#13735)) ([def1287](def1287)) * mint block rewards for 200K blocks at deployment ([#13537](#13537)) ([1a3a326](1a3a326)) * move check_circuit functionality from `TranslatorCircuitBuilder` into a `TranslatorCircuitChecker` ([#13761](#13761)) ([e4e7fca](e4e7fca)) * noir test flake addition ([#13663](#13663)) ([c302fdd](c302fdd)) * **noir-contracts:** update readme ([#13563](#13563)) ([42dfbbf](42dfbbf)) * **optimization:** Enable experimental ownership clone scheme by default (noir-lang/noir#8097) ([366d980](366d980)) * parse nop in SSA parser (noir-lang/noir#8141) ([366d980](366d980)) * ping luke instead of cody on bb benchmark regressions ([#13718](#13718)) ([2802f5a](2802f5a)) * playground improvements ([#13588](#13588)) ([4c3453b](4c3453b)) * **playground:** bump vite ([#13750](#13750)) ([7e72420](7e72420)) * Print `unsafe { ... }` around calls to Brillig from ACIR in AST (noir-lang/noir#8077) ([366d980](366d980)) * **public/avm:** from hints to the end of the world ([#13459](#13459)) ([a89de5d](a89de5d)) * redo typo PR by dizer-ti ([#13757](#13757)) ([075d3a8](075d3a8)) * release playground on s3. ([#13582](#13582)) ([2f9458a](2f9458a)) * **release:** give aztec release image sensible default entrypoint ([#13517](#13517)) ([a864c06](a864c06)) * remove circuit simulator ([#13689](#13689)) ([a00566c](a00566c)) * Remove code of the fake AVM recursive verifier ([#13614](#13614)) ([048dfed](048dfed)) * remove msm sorter ([#13668](#13668)) ([ff29d86](ff29d86)) * remove old terraform configs ([#13716](#13716)) ([49ac1db](49ac1db)), closes [#13651](#13651) * remove omit param from serialize derivation ([#13703](#13703)) ([6af5943](6af5943)) * remove pcs vk from vk ([#13638](#13638)) ([0e941ef](0e941ef)) * remove try_merge_only_changed_indices (noir-lang/noir#8142) ([366d980](366d980)) * replace relative paths to noir-protocol-circuits ([78c87b5](78c87b5)) * replace relative paths to noir-protocol-circuits ([4e53cd1](4e53cd1)) * replace relative paths to noir-protocol-circuits ([3a33c47](3a33c47)) * replace relative paths to noir-protocol-circuits ([eaaa85b](eaaa85b)) * replace relative paths to noir-protocol-circuits ([648a4fe](648a4fe)) * replace relative paths to noir-protocol-circuits ([b3f4745](b3f4745)) * replace relative paths to noir-protocol-circuits ([269b316](269b316)) * replace relative paths to noir-protocol-circuits ([1330efe](1330efe)) * replace relative paths to noir-protocol-circuits ([a739284](a739284)) * Sanity check block number from archiver before returning it ([#13631](#13631)) ([e459b2a](e459b2a)) * simpler `make_mutable` in `array_set` optimization (noir-lang/noir#8106) ([366d980](366d980)) * **spartan:** bump default kind resources ([#13747](#13747)) ([8ca828f](8ca828f)) * **ssa:** Additional unrolling and mem2reg passes (noir-lang/noir#8017) ([44c1347](44c1347)) * **ssa:** Remove IfElse pass basic SSA tests (noir-lang/noir#8058) ([44c1347](44c1347)) * **ssa:** Test terminator value constant folding and resolve cache for data bus (noir-lang/noir#8132) ([366d980](366d980)) * starknet feature flag ([#13681](#13681)) ([2b6c627](2b6c627)) * stop prover node swallowing start prover job errors ([#13676](#13676)) ([90c2b7b](90c2b7b)) * **test:** AST generation calibration test (noir-lang/noir#8054) ([44c1347](44c1347)) * update ACVM doc (noir-lang/noir#8004) ([366d980](366d980)) * Update docs snippet ([#13739](#13739)) ([7fb333a](7fb333a)) * update staking asset handler address ([#13566](#13566)) ([daa09f2](daa09f2)) * Updated contract addresses for alpha-testnet ([#13585](#13585)) ([9f0ff4a](9f0ff4a)) * Use chain monitor to sync system time in p2p tests ([#13632](#13632)) ([abbad4c](abbad4c)) * use public component key for pairing inputs ([#13705](#13705)) ([467166f](467166f)) * use PublicComponentKeys ([#13686](#13686)) ([68e4b6f](68e4b6f)) * use single extended viem client ([#13715](#13715)) ([cf637dc](cf637dc)) * validate block header hash in circuits ([#13094](#13094)) ([f19c182](f19c182)) * wasm proof verifying with native bb ([#13499](#13499)) ([1ad9249](1ad9249)) ### Documentation * fees tutorial ([#13357](#13357)) ([c580657](c580657)) * move snippets all versions ([#13650](#13650)) ([a2f5553](a2f5553)) * Update glossary & snippets ([#13661](#13661)) ([d55816d](d55816d)) --- This PR was generated with [Release Please](https://github.com/googleapis/release-please). See [documentation](https://github.com/googleapis/release-please#release-please). --------- Co-authored-by: AztecBot <[email protected]>

Apologies for the big PR. Lots of the stuff in here ended up being chicken and egg with wanting to do refactoring to make this process smoother/help debug tricky issues, and wanting to see that those refactorings actually make sense by the end. We no longer compute VKs on the fly in CIVC. This saves ~25% of computation. This is done throughout by consolidating IVC inputs into a single ivc-inputs.msgpack structure which supports passing bytecode, witness & vk information to the bb backend. Now attaches a name for each function, as well. Major features: - IVC inputs passed thru native and wasm are always passed a single file/buffer. This is encoded using msgpack and capture bytecode, witness, vk, and function name (which is now printed, but only properly passed by native) For native, the bincode and witnesses are gzipped, for WASM they are uncompressed. For actions such as gates or write_vk, the IVC inputs are used with same structure but witness and vk data can be blank. This has a bunch of implications, such as having to break away from the rigid API base class in bb cli (which overall doesn't feel worthwhile anyway as CIVC is fundamentally different than UH), having to string vk info along, etc. Other features: Debuggability: - Correct README.md instructions on WASM stack traces (give up on getting line numbers working :/) - clangd now properly shows all errors in a C++ file you're browsing, instead of only showing you the first error. Cleanup - small cleanup to acir tests, but still not testing new ivc flow there. Lightest weight test is ivc-integration in yarn-project - Get rid of --input_type in bb cli for CIVC. now implied always to be what was previously runtime_stack. Simplifies usages, other modes were unused. - more ignored linting in the clangd file. Maybe one day we can enforce the remaining as errors. - Clean up msgpack usage. Msgpack headers were leaking everywhere and it is a chunky library. - Consolidate with using msgpackr as our only typescript messagepack library Benches - use wasmtime helper in bb bootstrap. deduplicate code in bench. bench now honours NATIVE_PRESET, and if you do ``` export NATIVE_PRESET=op-count-time ./bootstrap.sh ./bootstrap.sh bench ``` you will get op count timings for our native ivc benches. --------- Co-authored-by: Copilot <[email protected]> Co-authored-by: thunkar <[email protected]> Co-authored-by: maramihali <[email protected]> Co-authored-by: ledwards2225 <[email protected]>

This PR fixes the issue in bb.js dependencies which is causing failures in noir-lang/noir#8506 This was broken in #13590 where the existing msgpack dependency was removed. Co-authored-by: AztecBot <[email protected]>

This PR fixes the issue in bb.js dependencies which is causing failures in noir-lang/noir#8506 This was broken in AztecProtocol/aztec-packages#13590 where the existing msgpack dependency was removed. Co-authored-by: AztecBot <[email protected]>

ludamad added 30 commits March 28, 2025 21:15

too surprising. we should consider removing bootstraps from hash

7a833ab

cache and bench cleanup, add e2e_ivc_bench to cpp/bootstrap.sh

33f2bd2

comment

1633a79

make a lib used just by main.cpp

7c864a8

rename to bb cli bench

dc29c03

e2e bench work

177bd39

dont tack on to a duplicated method...

ccbd8b1

more tweaks

1382798

Merge remote-tracking branch 'origin/master' into ad/e2e-ivc-bench

c91da61

libdeflate tweak

f6b2f9e

fixes

9aadff0

Merge remote-tracking branch 'origin/master' into ad/ivc-no-make-veri…

93d95ba

…fier-keys

allow calling build_nodejs_module

64cc855

Make sure bench script exits early on bad bb cli exit code

17b160f

fix main

fee32e2

benchmark work

d197965

zone scoped

081d5b4

fix execution of main.cpp in wasm

ce52bba

Merge remote-tracking branch 'origin/master' into ad/ivc-no-make-veri…

7591bf8

…fier-keys

port new main.cpp changes

cd266c1

add wasm calls to bench

11b6f24

benchmark both wasm and native

36c830c

ivc wasm benches

80c077e

fix msgpack usage

50d5b1a

make wasm normal again

98daf58

test fix

b81fb76

bench fix

1151d1b

Merge remote-tracking branch 'origin/master' into ad/ivc-verifier-bre…

12ebf59

…akdown

fix

54f82bb

fix

18cd4f4

ludamad added 2 commits April 17, 2025 14:44

typo

81cf33e

it was deduped, fix comment

a8a0ecc

ludamad commented Apr 17, 2025

View reviewed changes

ledwards2225 approved these changes Apr 17, 2025

View reviewed changes

ludamad added 2 commits April 17, 2025 15:57

try to move stuff around

29cf814

fix bench, I hope

d91561d

ludamad added this pull request to the merge queue Apr 17, 2025

github-merge-queue bot removed this pull request from the merge queue due to no response for status checks Apr 17, 2025

ludamad added this pull request to the merge queue Apr 17, 2025

ludamad removed this pull request from the merge queue due to a manual request Apr 17, 2025

ludamad merged commit 66a61ba into master Apr 17, 2025
7 checks passed

ludamad deleted the ad/ivc-verifier-breakdown branch April 17, 2025 20:39

AztecBot mentioned this pull request Apr 17, 2025

chore(master): release 0.86.0 #13579

Merged

ledwards2225 assigned ludamad Apr 17, 2025

ledwards2225 mentioned this pull request Apr 17, 2025

Add support for precomputed VKs to bb CIVC interface AztecProtocol/barretenberg#1247

Closed

TomAFrench mentioned this pull request May 15, 2025

fix: make msgpackr a full dependency of bb.js #14322

Merged


		pinned_civc_inputs_url="https://aztec-ci-artifacts.s3.us-east-2.amazonaws.com/protocol/bb-civc-inputs-v2.tar.gz"
		pinned_civc_inputs_url="https://aztec-ci-artifacts.s3.us-east-2.amazonaws.com/protocol/bb-civc-inputs-v3.tar.gz"

		@@ -30,11 +31,11 @@ if [ "${ISOLATE:-0}" -eq 1 ]; then
		-e NODE_OPTIONS="--no-warnings --experimental-vm-modules --loader @swc-node/register" \
		-e LOG_LEVEL \

feat: no longer recompute vk's in CIVC proofs #13590

feat: no longer recompute vk's in CIVC proofs #13590

Uh oh!

Conversation

ludamad commented Apr 16, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ludamad Apr 17, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ledwards2225 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ludamad Apr 17, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ludamad commented Apr 16, 2025 •

edited

Loading

ludamad Apr 17, 2025 •

edited

Loading

ludamad Apr 17, 2025 •

edited

Loading