aclnnMatmulReduceScatter

支持的产品型号

• Atlas A2训练系列产品/Atlas 800I A2推理产品。

说明： 使用该接口时，请确保驱动固件包和CANN包都为配套的8.0.RC2版本或者配套的更高版本，否则将会引发报错，比如BUS ERROR等。

接口原型

每个算子分为两段式接口，必须先调用“aclnnMatmulReduceScatterGetWorkspaceSize”接口获取计算所需workspace大小以及包含了算子计算流程的执行器，再调用“aclnnMatmulReduceScatter”接口执行计算。

• aclnnStatus aclnnMatmulReduceScatterGetWorkspaceSize(const aclTensor* x1, const aclTensor* x2, const aclTensor* bias, const char* group, const char* reduceOp, int64_t commTurn, int64_t stream_mode, const aclTensor* output, uint64_t* workspaceSize, aclOpExecutor** executor)
• aclnnStatus aclnnMatmulReduceScatter(void *workspace, uint64_t workspaceSize, aclOpExecutor *executor, aclrtStream stream)

功能描述

• 算子功能：完成mm + reduce_scatter_base计算。
• 计算公式：$$output=x1@x2$$

aclnnMatmulReduceScatterGetWorkspaceSize

• 参数说明：

  • x1（aclTensor*，计算输入）：Device侧的两维aclTensor，mm左矩阵。数据类型支持：FLOAT16、BFLOAT16，数据格式支持：ND。当前版本仅支持两维输入。
  • x2（aclTensor*，计算输入）：Device侧的两维aclTensor，mm右矩阵。数据类型支持：FLOAT16、BFLOAT16，数据格式支持：ND。当前版本仅支持两维输入。
  • bias（aclTensor*，计算输入）：Device侧的aclTensor。数据类型支持：FLOAT16、BFLOAT16。数据格式支持：ND。当前版本暂不支持bias输入为非0的场景。
  • group（char*，计算输入）：Host侧的char，标识通信域的字符串。数据类型支持：String。
  • reduceOp（char*，计算输入）：Host侧的char，reduce操作类型。数据类型支持：String。
  • commTurn（int64_t，计算输入）：Host侧的整型，通信数据切分数，即总数据量/单次通信量。数据类型支持：int64_t。当前版本仅支持输入0。
  • stream_mode（int64_t，计算输入）：Host侧的整型，acl流模式的枚举，当前只支持枚举值1，数据类型支持：int64_t。
  • output（aclTensor*，计算输出）：Device侧的aclTensor，计算+通信的结果。数据类型支持：FLOAT16、BFLOAT16，数据格式支持：ND。
  • workspaceSize（uint64_t*，出参）：Device侧的整型，返回需要在Device侧申请的workspace大小。
  • executor（aclOpExecutor**，出参）：Device侧的aclOpExecutor，返回op执行器，包含了算子计算流程。

• 返回值：

  返回aclnnStatus状态码，具体参见aclnn返回码。

aclnnMatmulReduceScatter

• 参数说明：

  • workspace（void*，入参）：在Device侧申请的workspace内存地址。
  • workspaceSize（uint64_t，入参）：在Device侧申请的workspace大小，由第一段接口aclnnAllGatherMatmulGetWorkspaceSize获取。
  • executor（aclOpExecutor*，入参）：op执行器，包含了算子计算流程。
  • stream（aclrtStream，入参）：指定执行任务的AscendCL stream流。

• 返回值：

  返回aclnnStatus状态码，具体参见aclnn返回码。

约束与限制

• 输入x1为2维，其维度为(m, k)：Atlas A2训练系列产品/Atlas 800I A2推理产品：m须为卡数rank_size的整数倍,。
• 输入x2必须是2维，其维度为(k, n)，轴满足mm算子入参要求，k轴相等，且k轴取值范围为[256, 65535)。bias暂不支持输入为非0的场景。
• 输出为2维，其维度为(m/rank_size, n), rank_size为卡数。
• x1、x2计算输入的数据类型要和output计算输出的数据类型一致。
• 只支持x2矩阵转置/不转置，x1矩阵支持不转置场景。
• Atlas A2训练系列产品/Atlas 800I A2推理产品支持2、4、8卡。

调用示例

#include <thread>
#include <iostream>
#include <vector>
#include "aclnnop/aclnn_matmul_reduce_scatter.h"

#define CHECK_RET(cond, return_expr) \
    do {                             \
        if (!(cond)) {               \
            return_expr;             \
        }                            \
    } while (0)

#define LOG_PRINT(message, ...)         \
    do {                                \
        printf(message, ##__VA_ARGS__); \
    } while(0)

constexpr int DEV_NUM = 8;

int64_t GetShapeSize(const std::vector<int64_t> &shape)
{
    int64_t shape_size = 1;
    for (auto i : shape) {
        shape_size *= i;
    }
    return shape_size;
}

template<typename T>
int CreateAclTensor(const std::vector<T> &hostData, const std::vector<int64_t> &shape, void **deviceAddr,
    aclDataType dataType, aclTensor **tensor)
{
    auto size = GetShapeSize(shape) * sizeof(T);
    auto ret = aclrtMalloc(deviceAddr, size, ACL_MEM_MALLOC_HUGE_FIRST);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("[ERROR] aclrtMalloc failed. ret: %d\n", ret); return ret);
    ret = aclrtMemcpy(*deviceAddr, size, hostData.data(), size, ACL_MEMCPY_HOST_TO_DEVICE);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("[ERROR] aclrtMemcpy failed. ret: %d\n", ret); return ret);
    std::vector<int64_t> strides(shape.size(), 1);
    for (int64_t i = shape.size() - 2; i >= 0; i--) {
        strides[i] = shape[i +1] * strides[i + 1];
    }
    *tensor = aclCreateTensor(shape.data(), shape.size(), dataType, strides.data(), 0, aclFormat::ACL_FORMAT_ND,
        shape.data(), shape.size(), *deviceAddr);
    return 0;
}

struct Args {
    int rankId;
    HcclComm hcclComm;
    aclrtStream stream;
  };

int launchOneThread_MmReduceScatter(Args &args)
{
    int ret = aclrtSetDevice(args.rankId);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("[ERROR] aclrtSetDevice failed. ret = %d \n", ret); return ret);

    char hcomName[128] = {0};
    ret = HcclGetCommName(args.hcclComm, hcomName);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("[ERROR] HcclGetCommName failed. ERROR: %d\n", ret); return -1);
    LOG_PRINT("[INFO] rank = %d, hcomName = %s, stream = %p\n", args.rankId, hcomName, args.stream);
    std::vector<int64_t> x1Shape = {1024, 256};
    std::vector<int64_t> x2Shape = {256, 512};
    std::vector<int64_t> biasShape = {512};
    std::vector<int64_t> outShape = {1024 / DEV_NUM, 512};
    void *x1DeviceAddr = nullptr;
    void *x2DeviceAddr = nullptr;
    void *biasDeviceAddr = nullptr;
    void *outDeviceAddr = nullptr;
    aclTensor *x1 = nullptr;
    aclTensor *x2 = nullptr;
    aclTensor *bias = nullptr;
    aclTensor *out = nullptr;

    int64_t commTurn = 0;
    int64_t stream_mode = 1;
    uint64_t workspaceSize = 0;
    aclOpExecutor *executor = nullptr;
    void *workspaceAddr = nullptr;

    long long x1ShapeSize = GetShapeSize(x1Shape);
    long long x2ShapeSize = GetShapeSize(x2Shape);
    long long biasShapeSize = GetShapeSize(biasShape);
    long long outShapeSize = GetShapeSize(outShape);

    std::vector<int16_t> x1HostData(x1ShapeSize, 0);
    std::vector<int16_t> x2HostData(x2ShapeSize, 0);
    std::vector<int16_t> biasHostData(biasShapeSize, 0);
    std::vector<int16_t> outHostData(outShapeSize, 0);
    // 创建tensor
    ret = CreateAclTensor(x1HostData, x1Shape, &x1DeviceAddr, aclDataType::ACL_FLOAT16, &x1);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(x2HostData, x2Shape, &x2DeviceAddr, aclDataType::ACL_FLOAT16, &x2);
    CHECK_RET(ret == ACL_SUCCESS, return ret);
    ret = CreateAclTensor(outHostData, outShape, &outDeviceAddr, aclDataType::ACL_FLOAT16, &out);
    CHECK_RET(ret == ACL_SUCCESS, return ret);

    // 调用第一阶段接口
    ret = aclnnMatmulReduceScatterGetWorkspaceSize(
        x1, x2, bias, hcomName, "sum", commTurn, stream_mode, out, &workspaceSize, &executor);
    CHECK_RET(ret == ACL_SUCCESS,
        LOG_PRINT("[ERROR] aclnnMatmulReduceScatterGetWorkspaceSize failed. ret = %d \n", ret); return ret);
    // 根据第一阶段接口计算出的workspaceSize申请device内存
    if (workspaceSize > 0) {
        ret = aclrtMalloc(&workspaceAddr, workspaceSize, ACL_MEM_MALLOC_HUGE_FIRST);
        CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("[ERROR] aclrtMalloc workspace failed. ret = %d \n", ret); return ret);
    }
    // 调用第二阶段接口
    ret = aclnnMatmulReduceScatter(workspaceAddr, workspaceSize, executor, args.stream);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("[ERROR] aclnnMatmulReduceScatter failed. ret = %d \n", ret); return ret);
    // （固定写法）同步等待任务执行结束
    ret = aclrtSynchronizeStreamWithTimeout(args.stream, 10000);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("[ERROR] aclrtSynchronizeStreamWithTimeout failed. ret = %d \n", ret);
        return ret);
    LOG_PRINT("[INFO] device_%d aclnnMatmulReduceScatter execute successfully.\n", args.rankId);
    // 释放device资源，需要根据具体API的接口定义修改
    if (x1 != nullptr) {
        aclDestroyTensor(x1);
    }
    if (x2 != nullptr) {
        aclDestroyTensor(x2);
    }
    if (bias != nullptr) {
        aclDestroyTensor(bias);
    }
    if (out != nullptr) {
        aclDestroyTensor(out);
    }
    if (x1DeviceAddr != nullptr) {
        aclrtFree(x1DeviceAddr);
    }
    if (x2DeviceAddr != nullptr) {
        aclrtFree(x2DeviceAddr);
    }
    if (biasDeviceAddr != nullptr) {
        aclrtFree(biasDeviceAddr);
    }
    if (outDeviceAddr != nullptr) {
        aclrtFree(outDeviceAddr);
    }
    if (workspaceSize > 0) {
        aclrtFree(workspaceAddr);
    }
    ret = aclrtDestroyStream(args.stream);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("[ERROR] aclrtDestroyStream failed. ret = %d \n", ret); return ret);
    ret = aclrtResetDevice(args.rankId);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("[ERROR] aclrtResetDevice failed. ret = %d \n", ret); return ret);
    return 0;
}

int main(int argc, char *argv[])
{
    int ret = aclInit(nullptr);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("[ERROR] aclInit failed. ret = %d \n", ret); return ret);
    aclrtStream stream[DEV_NUM];
    for (uint32_t rankId = 0; rankId < DEV_NUM; rankId++) {
        ret = aclrtSetDevice(rankId);
        CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("[ERROR] aclrtSetDevice failed. ret = %d \n", ret); return ret);
        ret = aclrtCreateStream(&stream[rankId]);
        CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("[ERROR] aclrtCreateStream failed. ret = %d \n", ret); return ret);
    }
    int32_t devices[DEV_NUM];
    for (int i = 0; i < DEV_NUM; i++) {
        devices[i] = i;
    }
    // 初始化集合通信域
    HcclComm comms[DEV_NUM];
    ret = HcclCommInitAll(DEV_NUM, devices, comms);
    CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("[ERROR] HcclCommInitAll failed. ret = %d \n", ret); return ret);

    Args args[DEV_NUM];
    // 启动多线程
    std::vector<std::unique_ptr<std::thread>> threads(DEV_NUM);
    for (uint32_t rankId = 0; rankId < DEV_NUM; rankId++) {
        args[rankId].rankId = rankId;
        args[rankId].hcclComm = comms[rankId];
        args[rankId].stream = stream[rankId];
        threads[rankId].reset(new(std::nothrow) std::thread(&launchOneThread_MmReduceScatter, std::ref(args[rankId])));    
    }
    for (uint32_t rankId = 0; rankId < DEV_NUM; rankId++) {
        threads[rankId]->join();
    }
    for (int i = 0; i < DEV_NUM; i++) {
        auto hcclRet = HcclCommDestroy(comms[i]);
        CHECK_RET(hcclRet == HCCL_SUCCESS, LOG_PRINT("[ERROR] HcclCommDestory failed. ret = %d \n", ret); return -1);
    }
    aclFinalize();
    return 0;
}