Skip to content

stdlib/shardView source ↗

shard

6 items · generated by scribe fromstdlib/shard.fd· synced 2026-11-15

stdlib/shard6 items

Contents

all_reduce

All-reduce: sum tensor across all ranks.

pub fn all_reduce(t: Tensor) -> Tensor {

Parameters:

  • t: Tensor

Returns: Tensor

Defined at line 10


broadcast

Broadcast tensor from rank 0 to all ranks.

pub fn broadcast(t: Tensor) -> Tensor {

Parameters:

  • t: Tensor

Returns: Tensor

Defined at line 15


all_gather

All-gather: concatenate tensors from all ranks.

pub fn all_gather(t: Tensor) -> Tensor {

Parameters:

  • t: Tensor

Returns: Tensor

Defined at line 20


scatter

Scatter: split tensor across ranks.

pub fn scatter(t: Tensor) -> Tensor {

Parameters:

  • t: Tensor

Returns: Tensor

Defined at line 25


send_tensor

Send tensor to another rank.

pub fn send_tensor(t: Tensor) -> void {

Parameters:

  • t: Tensor

Returns: void

Defined at line 30


recv_tensor

Receive tensor from another rank.

pub fn recv_tensor(shape: Vec<i64>) -> Tensor {

Parameters:

  • shape: Vec<i64>

Returns: Tensor

Defined at line 35


Generated by scribe, the Fides documentation generator