Struct DistributedContext 

pub struct DistributedContext { /* private fields */ }

A handle to the distributed runtime.

Implementations

impl DistributedContext

pub fn new(backend: Box<dyn DistributedBackend>) -> Self

Create a new distributed context with the given backend.

pub fn with_metrics( backend: Box<dyn DistributedBackend>, metrics: SharedMetrics, ) -> Self

Create a new distributed context with metrics enabled.

pub fn rank(&self) -> usize

Get the rank of this node.

pub fn world_size(&self) -> usize

Get the total number of nodes in the cluster.

pub async fn all_reduce(&self, buffer: &mut [u8], op: ReduceOp) -> Result<()>

Perform an all-reduce operation on the buffer.

After this call, all nodes will have the same values in their buffers. op controls whether the result is a sum or mean across nodes.

pub async fn barrier(&self) -> Result<()>

Synchronize all nodes at a barrier.

All nodes must call this method, and none will proceed until all have.

pub fn is_master(&self) -> bool

Check if this is the master node (rank 0).

pub fn metrics_snapshot(&self) -> Option<MetricsSnapshot>

Get metrics snapshot if enabled.