Struct AllGatherRequest 

pub struct AllGatherRequest<T: NcclReduceSupported> {
    pub send: GpuRef<T>,
    pub recv: GpuRef<T>,
    pub reply: Sender<Result<(), GpuError>>,
}

AllGather: each rank contributes send (length N) and writes a concatenated buffer of length N * world_size into recv.

Fields

send: GpuRef<T>

recv: GpuRef<T>

reply: Sender<Result<(), GpuError>>

Trait Implementations

impl<T: NcclReduceSupported> CollectiveDispatch for AllGatherRequest<T>

fn dtype_kind(&self) -> DispatchDType

fn device_id(&self) -> Option<u32>

fn dispatch(self: Box<Self>, ctx: &CollectiveDispatchCtx<'_>)