Crate monarch_rdma

Crate monarch_rdma 

Source

Re-exports§

pub use rdmaxcel_sys;
pub use backend::ibverbs::primitives::*;

Modules§

backend
RDMA backend implementations.
config
RDMA configuration attributes.
device_selection
PCI topology parsing and device discovery utilities for RDMA device selection.
efa
EFA (Elastic Fabric Adapter) specific RDMA operations.
local_memory
Local memory abstractions for RDMA operations.

Macros§

cu_check

Structs§

GetIbvActorRef
Serializable query for resolving the IbvManagerActor ref from a remote RdmaManagerActor. Only used in testing.
GetTcpActorRef
Serializable query for resolving the TcpManagerActor ref from a remote RdmaManagerActor.
RdmaManagerActor
RdmaOp
A single RDMA operation to be submitted to a backend.
RdmaRemoteBuffer
Lightweight handle representing a registered RDMA buffer.
ReleaseBuffer
Serializable release message for wire transport.

Enums§

RdmaLocalBackend
Backend handle returned by RdmaRemoteBuffer::choose_backend.
RdmaManagerMessage
Local-only messages for the RdmaManagerActor.
RdmaOpType
Type of RDMA operation.
RdmaTransportLevel
Transport level, ordered slowest to fastest.

Traits§

GetIbvActorRefClient
The custom client trait for this message type.
GetIbvActorRefHandler
The custom handler trait for this message type.
GetTcpActorRefClient
The custom client trait for this message type.
GetTcpActorRefHandler
The custom handler trait for this message type.
RdmaManagerMessageClient
The custom client trait for this message type.
RdmaManagerMessageHandler
The custom handler trait for this message type.
ReleaseBufferClient
The custom client trait for this message type.
ReleaseBufferHandler
The custom handler trait for this message type.

Functions§

get_rdmaxcel_error_message
Helper function to get detailed error messages from RDMAXCEL error codes
get_registered_cuda_segments
Get all segments that have been registered with MRs
is_cuda_available
Safely checks if CUDA is available on the system.
print_device_info
Print comprehensive RDMA device information for debugging (always prints).
print_device_info_if_debug_enabled
Print comprehensive RDMA device information for debugging. Controlled by MONARCH_DEBUG_RDMA environment variable.
rdma_supported
Whether any RDMA backend is available on this system.
register_segment_scanner
Register a segment scanner callback.
validate_execution_context
Utility to validate execution context.

Type Aliases§

SegmentScannerFn
Segment scanner callback type alias for convenience.