API

This section provides details of the library API

Communicator Functions

ncclResult_t ncclGetUniqueId(ncclUniqueId *uniqueId)

ncclResult_t ncclCommInitRank(ncclComm_t *comm, int nranks, ncclUniqueId commId, int rank)

ncclResult_t ncclCommInitAll(ncclComm_t *comm, int ndev, const int *devlist)

ncclResult_t ncclCommDestroy(ncclComm_t comm)

ncclResult_t ncclCommAbort(ncclComm_t comm)

ncclResult_t ncclCommCount(const ncclComm_t comm, int *count)

ncclResult_t ncclCommCuDevice(const ncclComm_t comm, int *device)

ncclResult_t ncclCommUserRank(const ncclComm_t comm, int *rank)

Collective Communication Operations

Collective communication operations must be called separately for each communicator in a communicator clique.

They return when operations have been enqueued on the hipstream.

Since they may perform inter-CPU synchronization, each call has to be done from a different thread or process, or need to use Group Semantics (see below).

ncclResult_t ncclReduce(const void *sendbuff, void *recvbuff, size_t count, ncclDataType_t datatype, ncclRedOp_t op, int root, ncclComm_t comm, hipStream_t stream)

ncclResult_t ncclBcast(void *buff, size_t count, ncclDataType_t datatype, int root, ncclComm_t comm, hipStream_t stream)

ncclResult_t ncclBroadcast(const void *sendbuff, void *recvbuff, size_t count, ncclDataType_t datatype, int root, ncclComm_t comm, hipStream_t stream)

ncclResult_t ncclAllReduce(const void *sendbuff, void *recvbuff, size_t count, ncclDataType_t datatype, ncclRedOp_t op, ncclComm_t comm, hipStream_t stream)

ncclResult_t ncclReduceScatter(const void *sendbuff, void *recvbuff, size_t recvcount, ncclDataType_t datatype, ncclRedOp_t op, ncclComm_t comm, hipStream_t stream)

ncclResult_t ncclAllGather(const void *sendbuff, void *recvbuff, size_t sendcount, ncclDataType_t datatype, ncclComm_t comm, hipStream_t stream)

Warning

doxygenfunction: Cannot find function “ncclSend” in doxygen xml output for project “RCCL 2.10.3 Documentation” from directory: /home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-rccl/checkouts/docs-5.0.2/docs/doxygen/docBin/xml

Warning

doxygenfunction: Cannot find function “ncclRecv” in doxygen xml output for project “RCCL 2.10.3 Documentation” from directory: /home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-rccl/checkouts/docs-5.0.2/docs/doxygen/docBin/xml

Warning

doxygenfunction: Cannot find function “ncclGather” in doxygen xml output for project “RCCL 2.10.3 Documentation” from directory: /home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-rccl/checkouts/docs-5.0.2/docs/doxygen/docBin/xml

Warning

doxygenfunction: Cannot find function “ncclScatter” in doxygen xml output for project “RCCL 2.10.3 Documentation” from directory: /home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-rccl/checkouts/docs-5.0.2/docs/doxygen/docBin/xml

Warning

doxygenfunction: Cannot find function “ncclAllToAll” in doxygen xml output for project “RCCL 2.10.3 Documentation” from directory: /home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-rccl/checkouts/docs-5.0.2/docs/doxygen/docBin/xml

Group Semantics

When managing multiple GPUs from a single thread, and since NCCL collective calls may perform inter-CPU synchronization, we need to “group” calls for different ranks/devices into a single call.

Grouping NCCL calls as being part of the same collective operation is done using ncclGroupStart and ncclGroupEnd. ncclGroupStart will enqueue all collective calls until the ncclGroupEnd call, which will wait for all calls to be complete. Note that for collective communication, ncclGroupEnd only guarantees that the operations are enqueued on the streams, not that the operation is effectively done.

Both collective communication and ncclCommInitRank can be used in conjunction of ncclGroupStart/ncclGroupEnd.

ncclResult_t ncclGroupStart()

ncclResult_t ncclGroupEnd()

Library Functions

ncclResult_t ncclGetVersion(int *version)

const char *ncclGetErrorString(ncclResult_t result)

Types

There are few data structures that are internal to the library. The pointer types to these structures are given below. The user would need to use these types to create handles and pass them between different library functions.

typedef struct ncclComm *ncclComm_t

struct ncclUniqueId

Enumerations

This section provides all the enumerations used.

enum ncclResult_t

Values:

enumerator ncclSuccess

enumerator ncclUnhandledCudaError

enumerator ncclSystemError

enumerator ncclInternalError

enumerator ncclInvalidArgument

enumerator ncclInvalidUsage

enumerator ncclNumResults

enum ncclRedOp_t

Values:

enumerator ncclSum

enumerator ncclProd

enumerator ncclMax

enumerator ncclMin

enumerator ncclNumOps

enum ncclDataType_t

Values:

enumerator ncclInt8

enumerator ncclChar

enumerator ncclUint8

enumerator ncclInt32

enumerator ncclInt

enumerator ncclUint32

enumerator ncclInt64

enumerator ncclUint64

enumerator ncclFloat16

enumerator ncclHalf

enumerator ncclFloat32

enumerator ncclFloat

enumerator ncclFloat64

enumerator ncclDouble

enumerator ncclBfloat16

enumerator ncclNumTypes