linear_parallel_demo.cpp

前置条件和编译命令请参见算子调用示例。

场景：基础场景。

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
#include <acl/acl.h>
#include <atb/atb_infer.h>
#include <iostream>
#include <unistd.h>
#include <sys/wait.h>

#include "demo_util.h"

void ExcuteImpl(atb::Operation *op, atb::VariantPack variantPack, atb::Context *context)
{
    uint64_t workspaceSize = 0;
    CHECK_STATUS(op->Setup(variantPack, workspaceSize, context));
    void *workspace = nullptr;
    if (workspaceSize > 0) {
        CHECK_STATUS(aclrtMalloc(&workspace, workspaceSize, ACL_MEM_MALLOC_HUGE_FIRST));
    }
    CHECK_STATUS(op->Execute(variantPack, (uint8_t *)workspace, workspaceSize, context));

    if (workspace) {
        CHECK_STATUS(aclrtFree(workspace));  // 销毁workspace
    }
}
void LinearParallelSample(int rank, int rankSize)
{
    int ret = aclInit(nullptr);
    // 设置每个进程对应的deviceId
    int deviceId = rank;
    CHECK_STATUS(aclrtSetDevice(deviceId));

    atb::Context *context = nullptr;
    CHECK_STATUS(atb::CreateContext(&context));
    aclrtStream stream = nullptr;
    CHECK_STATUS(aclrtCreateStream(&stream));
    context->SetExecuteStream(stream);

    atb::Tensor input = CreateTensorFromVector(
        context, stream, std::vector<float>(64, 2.0), aclDataType::ACL_FLOAT16, aclFormat::ACL_FORMAT_ND, {2, 32});
    atb::Tensor weight = CreateTensorFromVector(
        context, stream, std::vector<float>(64, 2.0), aclDataType::ACL_FLOAT16, aclFormat::ACL_FORMAT_ND, {32, 2});

    atb::Tensor output;
    output.desc.dtype = ACL_FLOAT16;
    output.desc.format = ACL_FORMAT_ND;
    output.desc.shape.dimNum = 2;
    output.desc.shape.dims[0] = 2;
    output.desc.shape.dims[1] = 2;
    output.dataSize = atb::Utils::GetTensorSize(output);
    CHECK_STATUS(aclrtMalloc(&output.deviceData, output.dataSize, ACL_MEM_MALLOC_HUGE_FIRST));

    atb::infer::LinearParallelParam param;
    param.transWeight = false;
    param.rank = rank;
    param.rankRoot = 0;
    param.rankSize = rankSize;
    param.backend = "hccl";
    atb::Operation *op = nullptr;
    CHECK_STATUS(atb::CreateOperation(param, &op));

    atb::VariantPack variantPack;
    variantPack.inTensors = {input, weight};
    variantPack.outTensors = {output};
    ExcuteImpl(op, variantPack, context);
    std::cout << "rank: " << rank << " executed END." << std::endl;

    // 资源释放
    CHECK_STATUS(atb::DestroyOperation(op));     // 销毁op对象
    CHECK_STATUS(aclrtDestroyStream(stream));    // 销毁stream
    CHECK_STATUS(atb::DestroyContext(context));  // 销毁context
    CHECK_STATUS(aclFinalize());
    std::cout << "demo excute success" << std::endl;
}

int main(int argc, const char *argv[])
{
    const int processCount = 2;
    for (int i = 0; i < processCount; i++) {
        pid_t pid = fork();
        // 子进程
        if (pid == 0) {
            LinearParallelSample(i, processCount);
            return 0;
        } else if (pid < 0) {
            std::cerr << "Failed to create process." << std::endl;
            return 1;
        }
    }
    // 父进程等待子进程执行完成
    for (int i = 0; i < processCount; ++i) {
        wait(NULL);
    }
    std::cout << "The communication operator is successfully executed. Parent process exit" << std::endl;
    return 0;
}
父主题： 算子调用示例（C++）