HcclSend

Applicability

Product	Supported
Atlas 350 Accelerator Card	√
Atlas A3 training product / Atlas A3 inference product	√
Atlas A2 training product / Atlas A2 inference product	√
Atlas 200I/500 A2 inference product	☓
Atlas inference product	☓
Atlas training product	√

For Atlas A2 training product / Atlas A2 inference product , only the Atlas 800T A2 training server, Atlas 900 A2 PoD cluster basic unit, and Atlas 200T A2 Box16 heterogeneous subrack are supported.

Function

Sends the data at the specified location on the current rank to the specified location on the destination rank.

Prototype

      
           HcclResult HcclSend(void* sendBuf, uint64_t count, HcclDataType dataType, uint32_t destRank, HcclComm comm, aclrtStream stream)

Parameters

Parameter	Input/Output	Description
sendBuf	Input	Address of the send buffer.
count	Input	Number of data records to be sent.
dataType	Input	Data type of the data to be sent, which is of the HcclDataType type. Atlas 350 Accelerator Card: The supported data types are int8, uint8, int16, uint16, int32, uint32, int64, uint64, float8-e5m2, float8-e4m3, float8-e8m0, hifloat8, float16, float32, float64, and bfp16. Atlas A3 training product / Atlas A3 inference product : The supported data types are int8, uint8, int16, uint16, int32, uint32, int64, uint64, float16, float32, float64, and bfp16. Atlas A2 training product / Atlas A2 inference product : The supported data types are int8, uint8, int16, uint16, int32, uint32, int64, uint64, float16, float32, float64, and bfp16. Atlas training product : The supported data types are int8, uint8, int16, uint16, int32, uint32, int64, uint64, float16, float32, and float64.
destRank	Input	ID of the destination rank in the communicator.
comm	Input	Communicator where the operation is performed.
stream	Input	Stream of the rank.

Returns

HcclResult: HCCL_SUCCESS on success, or else failure.

Constraints

The HcclSend and HcclRecv interfaces are invoked synchronously and must be used in pairs. That is, after a process calls the HcclSend interface, the process can call the next interface only after the corresponding HcclRecv interface receives data, as shown in the following figure.

Call Example

      
       
         
         
           void *sendBuf = nullptr;
void *recvBuf = nullptr;
uint64_t count = 8;
size_t mallocSize = count * sizeof(float);

// Initialize the communicator.
uint32_t rankSize = 8;
HcclComm hcclComm;
HcclCommInitRootInfo(rankSize, &rootInfo, deviceId, &hcclComm);

// Create a task flow.
aclrtStream stream;
aclrtCreateStream(&stream);

// Perform the Send/Recv operation. Devices 0, 2, 4, and 6 send data, and devices 1, 3, 5, and 7 receive data.
// The HcclSend and HcclRecv APIs are called synchronously and must be used in pairs.
if (deviceId % 2 == 0) {
    // Allocate the device memory for storing the input data.
    aclrtMalloc(&sendBuf, mallocSize, ACL_MEM_MALLOC_HUGE_ONLY);
    // Initialize the input data.
    aclrtMemcpy(sendBuf, mallocSize, hostBuf, mallocSize, ACL_MEMCPY_HOST_TO_DEVICE);
    // Perform the Send operation.
    HcclSend(sendBuf, count, HCCL_DATA_TYPE_FP32, deviceId + 1, hcclComm, stream);
} else {
    // Allocate the device memory for receiving data.
    aclrtMalloc(&recvBuf, mallocSize, ACL_MEM_MALLOC_HUGE_ONLY);
    // Perform the Recv operation.
    HcclRecv(recvBuf, count, HCCL_DATA_TYPE_FP32, deviceId - 1, hcclComm, stream);
}

// Wait until the collective communication task in the task flow is complete.
aclrtSynchronizeStream(stream);

// Free resources.
aclrtFree(sendBuf);          // Free the device memory.
aclrtFree(recvBuf);          // Free the device memory.
aclrtDestroyStream(stream);  // Destroy the task flow.
HcclCommDestroy(hcclComm);   // Destroy the communicator.

          

        

      
     

Parent topic: Point-to-Point Communication