bb::scalar_multiplication::MSM_fast< Curve > Class Template Reference

#include <scalar_multiplication_fast.hpp>

Public Types
using	Element = typename Curve::Element
using	ScalarField = typename Curve::ScalarField
using	AffineElement = typename Curve::AffineElement

Static Public Member Functions
static AffineElement	msm (std::span< const AffineElement > points, PolynomialSpan< const ScalarField > scalars, bool handle_edge_cases=false, bool dedup_hint=false) noexcept
	Single MSM_fast convenience wrapper — returns the result as an AffineElement.
static std::vector< AffineElement >	batch_multi_scalar_mul (std::span< const AffineElement > points, std::span< PolynomialSpan< ScalarField > > scalars, bool handle_edge_cases=true, std::span< const uint8_t > dedup_hints={}) noexcept
	Batch driver for multiple MSMs. Returns one AffineElement per input MSM_fast.

Detailed Description

template<typename Curve>
class bb::scalar_multiplication::MSM_fast< Curve >

Definition at line 128 of file scalar_multiplication_fast.hpp.

Member Typedef Documentation

AffineElement

template<typename Curve >

using bb::scalar_multiplication::MSM_fast< Curve >::AffineElement = typename Curve::AffineElement

Definition at line 132 of file scalar_multiplication_fast.hpp.

Element

template<typename Curve >

using bb::scalar_multiplication::MSM_fast< Curve >::Element = typename Curve::Element

Definition at line 130 of file scalar_multiplication_fast.hpp.

ScalarField

template<typename Curve >

using bb::scalar_multiplication::MSM_fast< Curve >::ScalarField = typename Curve::ScalarField

Definition at line 131 of file scalar_multiplication_fast.hpp.

Member Function Documentation

batch_multi_scalar_mul()

template<typename Curve >

std::vector< typename Curve::AffineElement > MSM_fast::batch_multi_scalar_mul ( std::span< const AffineElement >  points, std::span< PolynomialSpan< ScalarField > >  scalars, bool  handle_edge_cases = true, std::span< const uint8_t >  dedup_hints = {}  )

staticnoexcept

Batch driver for multiple MSMs. Returns one AffineElement per input MSM_fast.

Every MSM_fast in the batch shares a single contiguous point set (the SRS / GLV table); each MSM_fast picks its own range via scalars[m].start_index and scalars[m].size(). MSM_fast m computes Σ_i scalars[m][i] * points[scalars[m].start_index + i].

Independent MSMs run sequentially (each MSM_fast is itself round-parallel internally). This matches the legacy interface but the parallelisation strategy is different: the legacy implementation work-balanced points across threads spanning multiple MSMs; round-parallel parallelises within each MSM_fast, so one MSM_fast at a time uses the full thread pool. For the typical chonk workload (commit batches of polys of size 2^20), this is faster because per-MSM_fast threading dominates over inter-MSM_fast stealing.

Parameters

points	Shared point set (SRS prefix). Every MSM_fast indexes into this span.
scalars	Per-MSM_fast scalars carrying the start offset into points.
dedup_hints	Optional per-MSM_fast dedup opt-ins (parallel to scalars): a non-zero entry opts that MSM_fast's input scalars into the duplicate-cluster pre-pass. Empty span means no dedup anywhere.

Definition at line 231 of file pippenger_batched.hpp.

msm()

template<typename Curve >

Curve::AffineElement bb::scalar_multiplication::MSM_fast< Curve >::msm ( std::span< const AffineElement >  points, PolynomialSpan< const ScalarField >  scalars, bool  handle_edge_cases = false, bool  dedup_hint = false  )

staticnoexcept

Single MSM_fast convenience wrapper — returns the result as an AffineElement.

Parameters

handle_edge_cases	false (default): fast affine round-parallel path. true: Jacobian fast path (handles edge cases).
dedup_hint	When true, opts this MSM_fast into the input-scalar dedup pre-pass.

Definition at line 2908 of file scalar_multiplication_fast.cpp.

---

The documentation for this class was generated from the following files:

• src/barretenberg/ecc/scalar_multiplication/scalar_multiplication_fast.hpp
• src/barretenberg/ecc/scalar_multiplication/pippenger_batched.hpp
• src/barretenberg/ecc/scalar_multiplication/scalar_multiplication_fast.cpp