Struct MmapGradientAccumulator 

pub struct MmapGradientAccumulator { /* private fields */ }

Memory-mapped gradient accumulator with fine-grained locking.

Allows multiple threads to accumulate gradients concurrently with minimal contention. Uses reader-writer locks at a configurable granularity.

Implementations

impl MmapGradientAccumulator

pub fn new(path: &Path, d_embed: usize, max_nodes: usize) -> Result<Self>

Create a new memory-mapped gradient accumulator.

Arguments

• path - Path to the gradient file
• d_embed - Embedding dimension
• max_nodes - Maximum number of nodes

Returns

A new MmapGradientAccumulator instance

pub fn grad_offset(&self, node_id: u64) -> usize

Calculate the byte offset for a node’s gradient.

Arguments

• node_id - Node identifier

Returns

Byte offset in the gradient file

pub fn accumulate(&self, node_id: u64, grad: &[f32])

Accumulate gradients for a specific node.

Arguments

• node_id - Node identifier
• grad - Gradient vector to accumulate

Panics

Panics if grad length doesn’t match d_embed

pub fn apply(&mut self, learning_rate: f32, embeddings: &mut MmapManager)

Apply accumulated gradients to embeddings and zero out gradients.

Arguments

• learning_rate - Learning rate for gradient descent
• embeddings - Embedding manager to update

pub fn zero_grad(&mut self)

Zero out all accumulated gradients.

pub fn get_grad(&self, node_id: u64) -> &[f32]

Get a read-only reference to a node’s accumulated gradient.

Arguments

• node_id - Node identifier

Returns

Slice containing the gradient vector

pub fn d_embed(&self) -> usize

Get the embedding dimension.

pub fn n_nodes(&self) -> usize

Get the number of nodes.

Trait Implementations

impl Drop for MmapGradientAccumulator

fn drop(&mut self)

Executes the destructor for this type.

impl Send for MmapGradientAccumulator

impl Sync for MmapGradientAccumulator