Cl/mum merge by charlielye · Pull Request #4 · AztecProtocol/barretenberg

charlielye · 2022-12-14T19:10:22Z

No description provided.

suyash67 · 2022-12-15T12:25:12Z

@@ -0,0 +1,294 @@
+/*


Why two blake3s and blake3s_full? Pls keep just one.

TODO: as a separate PR after this one's merged.

suyash67 · 2022-12-15T12:43:18Z

@@ -18,7 +18,6 @@ namespace {
 auto& engine = numeric::random::get_debug_engine();


Can we get rid of srs as a dependency of ecc? Because its used only in the tests.

TODO: do this as a separate PR, one idea can be to filter out io as its own module for read-write of transcripts. That will help us get rid of the circular dependency.

suyash67 · 2022-12-15T12:56:45Z

 }

-WASM_EXPORT void prover_process_queue(waffle::TurboProver* prover)
+typedef std::conditional_t<waffle::SYSTEM_COMPOSER == waffle::TURBO, waffle::TurboProver, waffle::UltraProver>


This prover should not be a compile time but runtime declaration. Use polymorphism (assign someone, should be done in a fresh PR)

suyash67 · 2022-12-15T19:48:41Z


-    monomials_[0] = barretenberg::g1::affine_one;
+    size_t index = 0;
+    if (!is_lagrange) {


Note to self: this conditional logic to handle the generator $[1]_1 \in \mathbb{G}_1$ creates the need for is_lagrange variable. If we can add $[1]_1$ to the first monomial transcript, we can get rid of this is_lagrange. Do this as a fresh PR.

suyash67 · 2022-12-15T20:13:57Z


    circuit_proving_key->recursive_proof_public_input_indices =
        std::vector<uint32_t>(recursive_proof_public_input_indices.begin(), recursive_proof_public_input_indices.end());
+    // What does this line do exactly?


Note to self: contains_recursive_proof was removed from the proving key iirc. Check this.

suyash67 · 2022-12-15T20:40:11Z

 }

-WASM_EXPORT size_t prover_get_circuit_size(waffle::TurboProver* prover)
+WASM_EXPORT size_t prover_get_circuit_size(WasmProver* prover)


Longer term, we want to use only the unrolled prover, so we should get rid of the linearised prover to reduce some code complexity.

suyash67 · 2022-12-15T21:16:57Z

+        barretenberg::polynomial& w_4_lagrange = key->polynomial_cache.get(wire_tag + "_lagrange");
+        // barretenberg::polynomial& wire_fft = key->polynomial_cache.get(wire_tag + "_fft");
+        barretenberg::polynomial w_4(key->n);
+        // barretenberg::polynomial_arithmetic::copy_polynomial(&wire[0], &wire_fft[0], n, n);


TODO: get rid of commented out parts.

suyash67 · 2022-12-15T21:32:13Z

@@ -49,17 +49,30 @@ void ifft_with_constant(fr* coeffs, const evaluation_domain& domain, const fr& v
 void coset_ifft(fr* coeffs, const evaluation_domain& domain);
 void coset_ifft(std::vector<fr*> coeffs, const evaluation_domain& domain);

+void partial_fft_serial_inner(fr* coeffs,


TODO: check if these functions are tested.

suyash67 · 2022-12-15T21:34:17Z

@@ -254,8 +255,8 @@ bool verify_proof(waffle::plonk_proof const& proof)
    UnrolledVerifier verifier(verification_key,
                              Composer::create_unrolled_manifest(verification_key->num_public_inputs));

-    std::unique_ptr<waffle::KateCommitmentScheme<waffle::unrolled_turbo_settings>> kate_commitment_scheme =
-        std::make_unique<waffle::KateCommitmentScheme<waffle::unrolled_turbo_settings>>();
+    std::unique_ptr<waffle::KateCommitmentScheme<waffle::unrolled_ultra_settings>> kate_commitment_scheme =


TODO (future): make this composer-agnostic.

suyash67 · 2022-12-15T21:41:48Z

@@ -0,0 +1,14 @@
+#!/bin/zsh


Do we need this? (ask Cody)

suyash67 · 2022-12-16T06:15:39Z

@@ -13,14 +14,14 @@ using namespace plonk;
 namespace {
 auto& engine = numeric::random::get_debug_engine();
 }
-typedef stdlib::packed_byte_array<waffle::TurboComposer> packed_byte_array;
-typedef stdlib::byte_array<waffle::TurboComposer> byte_array;
+typedef stdlib::packed_byte_array<waffle::UltraComposer> packed_byte_array;


TODO: use Composer from types.hpp.

suyash67 · 2022-12-16T06:19:02Z

-         : typeid(composer) == typeid(waffle::TurboComposer) ? "TurboPlonk" : "NULLPlonk")
+    (typeid(composer) == typeid(waffle::StandardComposer) ? "StandardPlonk"                                            \
+     : typeid(composer) == typeid(waffle::TurboComposer)  ? "TurboPlonk"                                               \
+                                                          : "NULLPlonk")


TODO: why NULLPlonk?

* Fix to bootstrap.sh after move into submodule (#1) * Updated bootstrap now that bberg may be a submodule, and now that cpp subdir exists * fix gitignore, wasnt properly ignoring cpp/build* * fix the barretenberg cpp formatting pre-commit hook now that barretenberg iss its own repo and may be a submodule * Contrived change. * Revert "Contrived change." This reverts commit 68aff9c. * Revert "Revert "Contrived change."" This reverts commit 3eee2fd. * Introduce cpp to paths. * More cpp path. * Cl/mum merge (#4) * Initial merge. Still has conflicts. * Tidied up bigfield and biggroup * Builds. * Order constants. * Fix. * Make more tests pass. * Update circuit change detection stuff to be local to tests. * Include header. * Cleanup and settings. * Path updates. * Path updates. Co-authored-by: Rumata888 <isennovskiy@gmail.com> Co-authored-by: David Banks <47112877+dbanks12@users.noreply.github.com> Co-authored-by: Rumata888 <isennovskiy@gmail.com>

* Fix to bootstrap.sh after move into submodule (AztecProtocol/barretenberg#1) * Updated bootstrap now that bberg may be a submodule, and now that cpp subdir exists * fix gitignore, wasnt properly ignoring cpp/build* * fix the barretenberg cpp formatting pre-commit hook now that barretenberg iss its own repo and may be a submodule * Contrived change. * Revert "Contrived change." This reverts commit f01a24e76362ed98875b7ad027ccce32802f3584. * Revert "Revert "Contrived change."" This reverts commit e500666fdcb9fb7d684a5fb717268fd0c6c88aea. * Introduce cpp to paths. * More cpp path. * Cl/mum merge (AztecProtocol/barretenberg#4) * Initial merge. Still has conflicts. * Tidied up bigfield and biggroup * Builds. * Order constants. * Fix. * Make more tests pass. * Update circuit change detection stuff to be local to tests. * Include header. * Cleanup and settings. * Path updates. * Path updates. Co-authored-by: Rumata888 <isennovskiy@gmail.com> Co-authored-by: David Banks <47112877+dbanks12@users.noreply.github.com> Co-authored-by: Rumata888 <isennovskiy@gmail.com>

* Fix to bootstrap.sh after move into submodule (AztecProtocol/barretenberg#1) * Updated bootstrap now that bberg may be a submodule, and now that cpp subdir exists * fix gitignore, wasnt properly ignoring cpp/build* * fix the barretenberg cpp formatting pre-commit hook now that barretenberg iss its own repo and may be a submodule * Contrived change. * Revert "Contrived change." This reverts commit a474f39b35fd024cc8be65dd8668124913da909e. * Revert "Revert "Contrived change."" This reverts commit 7a2ea1bcf35f1fa8b6d42bed7cf36d112ccafd45. * Introduce cpp to paths. * More cpp path. * Cl/mum merge (AztecProtocol/barretenberg#4) * Initial merge. Still has conflicts. * Tidied up bigfield and biggroup * Builds. * Order constants. * Fix. * Make more tests pass. * Update circuit change detection stuff to be local to tests. * Include header. * Cleanup and settings. * Path updates. * Path updates. Co-authored-by: Rumata888 <isennovskiy@gmail.com> Co-authored-by: David Banks <47112877+dbanks12@users.noreply.github.com> Co-authored-by: Rumata888 <isennovskiy@gmail.com>

Overview: If you are in a scenario where you have a failing call to check_circuit and wish to get more information out of it than just the gate index, you can use this feature to get a stack trace, see example below. Usage instructions: - On ubuntu (or our mainframe accounts) use `sudo apt-get install libdw-dev` to support trace printing - Use `cmake --preset clang16-dbg-fast-circuit-check-traces` and `cmake --build --preset clang16-dbg-fast-circuit-check-traces` to enable the backward-cpp dependency through the CHECK_CIRCUIT_STACKTRACES CMake variable. - Run any case where you have a failing check_circuit call, you will now have a stack trace illuminating where this constraint was added in code. Caveats: - This works best for code that is not overly generic, i.e. where just the sequence of function calls carries a lot of information. It is possible to tag extra data along with the stack trace, this can be done as a followup, please leave feedback if desired. - There are certain functions like `assert_equals` that can cause gates that occur _before_ them to fail. If this would be useful to automatically report, please leave feedback. Example: ``` [ RUN ] standard_circuit_constructor.test_check_circuit_broken Stack trace (most recent call last): #4 Source "_deps/gtest-src/googletest/src/gtest.cc", line 2845, in Run 2842: if (!Test::HasFatalFailure() && !Test::IsSkipped()) { 2843: // This doesn't throw as all user code that can throw are wrapped into 2844: // exception handling code. >2845: test->Run(); 2846: } 2847: 2848: if (test != nullptr) { #3 Source "_deps/gtest-src/googletest/src/gtest.cc", line 2696, in Run 2693: // GTEST_SKIP(). 2694: if (!HasFatalFailure() && !IsSkipped()) { 2695: impl->os_stack_trace_getter()->UponLeavingGTest(); >2696: internal::HandleExceptionsInMethodIfSupported(this, &Test::TestBody, 2697: "the test body"); 2698: } #2 | Source "_deps/gtest-src/googletest/src/gtest.cc", line 2657, in HandleSehExceptionsInMethodIfSupported<testing::Test, void> | 2655: #if GTEST_HAS_EXCEPTIONS | 2656: try { | >2657: return HandleSehExceptionsInMethodIfSupported(object, method, location); | 2658: } catch (const AssertionException&) { // NOLINT | 2659: // This failure was reported already. Source "_deps/gtest-src/googletest/src/gtest.cc", line 2621, in HandleExceptionsInMethodIfSupported<testing::Test, void> 2618: } 2619: #else 2620: (void)location; >2621: return (object->*method)(); 2622: #endif // GTEST_HAS_SEH 2623: } #1 Source "/mnt/user-data/adam/aztec-packages/barretenberg/cpp/src/barretenberg/circuit_checker/standard_circuit_builder.test.cpp", line 464, in TestBody 461: uint32_t d_idx = circuit_constructor.add_variable(d); 462: circuit_constructor.create_add_gate({ a_idx, b_idx, c_idx, fr::one(), fr::one(), fr::neg_one(), fr::zero() }); 463: > 464: circuit_constructor.create_add_gate({ d_idx, c_idx, a_idx, fr::one(), fr::neg_one(), fr::neg_one(), fr::zero() }); 465: 466: bool result = CircuitChecker::check(circuit_constructor); 467: EXPECT_EQ(result, false); #0 Source "/mnt/user-data/adam/aztec-packages/barretenberg/cpp/src/barretenberg/stdlib_circuit_builders/standard_circuit_builder.cpp", line 22, in create_add_gate 19: { 20: this->assert_valid_variables({ in.a, in.b, in.c }); 21: > 22: blocks.arithmetic.populate_wires(in.a, in.b, in.c); 23: blocks.arithmetic.q_m().emplace_back(FF::zero()); 24: blocks.arithmetic.q_1().emplace_back(in.a_scaling); 25: blocks.arithmetic.q_2().emplace_back(in.b_scaling); gate number4 ```

BEGIN_COMMIT_OVERRIDE chore: cycle group cleanup #3 (#16902) fix: improve polynomial structuring in translator and fix bug (#16941) refactor(bb): thread chunk for better work distribution (#16966) fix(bb): shell injection vectors (#16958) chore: cycle group cleanup #4 (#16937) fix: use consistent default build dir in benchmark_remote.sh (#17047) fix: misuse of slice() in cycle_scalar (#16982) vks update docs: improve some Protogalaxy and Translator docs (#17035) fix: edge case in the ECCVM related to splitting scalars (#16816) update vks hash chore: return of the instance (#17043) a second attempt at VK update fix: chmod +x test_civc_standalone_vks_havent_changed.sh fix avm build and update vks chore: Gemini high degree test (#17055) fix: update-vk refactor(bb): Optimize interleaved polynomial batching in Gemini with parallel chunking (#17049) END_COMMIT_OVERRIDE

charlielye and others added 4 commits December 14, 2022 15:05

Initial merge. Still has conflicts.

03e5fd0

Tidied up bigfield and biggroup

9008466

Builds.

cf96f3a

Order constants.

86faa4f

charlielye changed the base branch from master to aztec3 December 14, 2022 19:11

charlielye added 4 commits December 14, 2022 19:12

Fix.

db258f2

Make more tests pass.

fc04c87

Update circuit change detection stuff to be local to tests.

faa75aa

Include header.

add2948

suyash67 reviewed Dec 15, 2022

View reviewed changes

Comment thread cpp/CMakeLists.txt Outdated

suyash67 reviewed Dec 15, 2022

View reviewed changes

Comment thread cpp/dockerfiles/Dockerfile.arm64-linux-gcc Outdated

suyash67 reviewed Dec 15, 2022

View reviewed changes

charlielye added 3 commits December 15, 2022 14:09

Cleanup and settings.

7c4e001

Path updates.

75a8fa5

Path updates.

b3c3637

suyash67 reviewed Dec 15, 2022

View reviewed changes

Comment thread cpp/ultra_dev_tests.sh

@@ -0,0 +1,14 @@

#!/bin/zsh

suyash67 Dec 15, 2022

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do we need this? (ask Cody)

suyash67 reviewed Dec 16, 2022

View reviewed changes

Comment thread cpp/src/aztec/stdlib/encryption/aes128/aes128.cpp

charlielye merged commit 1fd4749 into aztec3 Dec 16, 2022

charlielye deleted the cl/mum_merge branch December 16, 2022 11:23

ledwards2225 mentioned this pull request Jan 3, 2024

Memory issue with ACIR + Goblin #819

Closed

ludamad mentioned this pull request Mar 26, 2024

MSAN: client_ivc_tests #926

Closed

		@@ -18,7 +18,6 @@ namespace {
		auto& engine = numeric::random::get_debug_engine();

Conversation

charlielye commented Dec 14, 2022

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

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!

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!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants