Python multiprocessing.get_context方法代碼示例

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def test_multi_client():
    ip_config = open("rpc_ip_config_mul_client.txt", "w")
    ip_addr = get_local_usable_addr()
    ip_config.write('%s 1\n' % ip_addr)
    ip_config.close()
    ctx = mp.get_context('spawn')
    pserver = ctx.Process(target=start_server, args=(10, "rpc_ip_config_mul_client.txt"))
    pclient_list = []
    for i in range(10):
        pclient = ctx.Process(target=start_client, args=("rpc_ip_config_mul_client.txt",))
        pclient_list.append(pclient)
    pserver.start()
    time.sleep(1)
    for i in range(10):
        pclient_list[i].start()
    for i in range(10):
        pclient_list[i].join()
    pserver.join() 

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def test_kernel_error_checking():
    # Running tests that may throw exceptions out of worker threads will stop CI testing
    # if not run in a separate process (with its own address space for CUDA compatibility).
    try:
        mpctx = mp.get_context('spawn')
    except:
        print('SKIP: python%s.%s lacks the required process fork-exec support ... ' %
              sys.version_info[0:2], file=sys.stderr, end='')
    else:
        with discard_stderr():
            for f in [kernel_error_check_imperative, kernel_error_check_symbolic]:
                p = mpctx.Process(target=f)
                p.start()
                p.join()
                assert p.exitcode != 0,\
                    "Expected a synchronous kernel error from %s(), none seen." % f.__name__ 

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def test_kv_store():
    # start 10 server and 10 client
    ip_config = open("kv_ip_config.txt", "w")
    ip_addr = get_local_usable_addr()
    ip_config.write('%s 10\n' % ip_addr)
    ip_config.close()
    ctx = mp.get_context('spawn')
    pserver_list = []
    pclient_list = []
    for i in range(10):
        pserver = ctx.Process(target=start_server, args=(i, 10))
        pserver.start()
        pserver_list.append(pserver)
    time.sleep(2)
    for i in range(10):
        pclient = ctx.Process(target=start_client, args=(10,))
        pclient.start()
        pclient_list.append(pclient)
    for i in range(10):
        pclient_list[i].join()
    for i in range(10):
        pserver_list[i].join() 

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def __init__(self, history_cls, history_args):
        super().__init__()

        self._last_needed_feed_count = 0
        self.results = {}
        self.pending_counts = {}

        # Make sure to use 'spawn' and not 'fork' to allow shared CUDA tensors
        # on linux
        ctx = mp.get_context('spawn')
        self.close_event = ctx.Event()
        self.qevent = ctx.Event()
        # Queue for requests, such as getting training data
        self.request_queue = ctx.Queue(10)
        # Queue for updates like new acting samples and priority updates
        self.update_queue = ctx.Queue(10)
        # Queue for sending back request results
        self.result_queue = ctx.Queue()

        self._process = ctx.Process(
            target=self.run,
            args=(history_cls, cloudpickle.dumps(history_args)))

        self._process.start() 

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def start(self):
        """
        Launch DagFileProcessorManager processor and start DAG parsing loop in manager.
        """
        mp_start_method = self._get_multiprocessing_start_method()
        context = multiprocessing.get_context(mp_start_method)

        self._parent_signal_conn, child_signal_conn = context.Pipe()
        self._process = context.Process(
            target=type(self)._run_processor_manager,
            args=(
                self._dag_directory,
                self._max_runs,
                # getattr prevents error while pickling an instance method.
                getattr(self, "_processor_factory"),
                self._processor_timeout,
                child_signal_conn,
                self._dag_ids,
                self._pickle_dags,
                self._async_mode
            )
        )
        self._process.start()

        self.log.info("Launched DagFileProcessorManager with pid: %s", self._process.pid) 

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def start(self):
        """
        Launch the process and start processing the DAG.
        """
        start_method = self._get_multiprocessing_start_method()
        context = multiprocessing.get_context(start_method)

        self._parent_channel, _child_channel = context.Pipe()
        self._process = context.Process(
            target=type(self)._run_file_processor,
            args=(
                _child_channel,
                self.file_path,
                self._pickle_dags,
                self._dag_ids,
                "DagFileProcessor{}".format(self._instance_id),
                self._failure_callback_requests
            ),
            name="DagFileProcessor{}-Process".format(self._instance_id)
        )
        self._start_time = timezone.utcnow()
        self._process.start() 

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def track_gpu_utils(device_ids: List[int],
                    interval: float = 0.05,
                    ) -> Generator[List[float], None, None]:
    # Spawn a worker.
    ctx = mp.get_context('spawn')
    conn, conn_worker = ctx.Pipe(duplex=True)
    p = ctx.Process(target=_worker, args=(device_ids, interval, conn_worker))
    p.start()
    conn.recv()

    # GPU% will be filled to this.
    gpu_utils: List[float] = []
    yield gpu_utils

    # Stop the worker and receive the timeline.
    conn.send(None)
    gpu_timeline = conn.recv()
    p.join()

    # Fill the GPU%.
    if gpu_timeline:
        gpu_utils.extend(sum(t)/len(t)/100 for t in zip(*gpu_timeline))
    else:
        gpu_utils.extend(0.0 for _ in device_ids) 

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def test_after_fork(self):
        writer = datasets.PickleDatasetWriter(self.io)
        writer.write(1)
        writer.flush()

        reader = ReaderMock(self.io)
        # Assign to avoid destruction of the instance
        # before creation a child process
        dataset = datasets.PickleDataset(reader)

        assert reader.n_hook_called == 0
        ctx = multiprocessing.get_context('fork')
        p = ctx.Process()
        p.start()
        p.join()
        assert reader.n_hook_called == 1
        assert reader.last_caller_pid == p.pid

        # Touch to suppress "unused variable' warning
        del dataset 

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def init_mp_pool(reset=False):
    """Necessary because at import time, cfg might be uninitialized"""
    global _mp_pool
    if _mp_pool and not reset:
        return _mp_pool

    cfg.CONFIG_MODIFIED = False
    if _mp_pool and reset:
        _mp_pool.terminate()
        _mp_pool = None

    if cfg.PARAMS['use_mp_spawn']:
        mp = multiprocessing.get_context('spawn')
    else:
        mp = multiprocessing

    cfg_contents = cfg.pack_config()
    global_lock = mp.Manager().Lock()

    mpp = cfg.PARAMS['mp_processes']
    _mp_pool = mp.Pool(mpp, initializer=_init_pool_globals,
                       initargs=(cfg_contents, global_lock))
    return _mp_pool 

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def test_set_get(self):
        multiprocessing.set_forkserver_preload(PRELOAD)
        count = 0
        old_method = multiprocessing.get_start_method()
        try:
            for method in ('fork', 'spawn', 'forkserver'):
                try:
                    multiprocessing.set_start_method(method, force=True)
                except ValueError:
                    continue
                self.assertEqual(multiprocessing.get_start_method(), method)
                ctx = multiprocessing.get_context()
                self.assertEqual(ctx.get_start_method(), method)
                self.assertTrue(type(ctx).__name__.lower().startswith(method))
                self.assertTrue(
                    ctx.Process.__name__.lower().startswith(method))
                self.check_context(multiprocessing)
                count += 1
        finally:
            multiprocessing.set_start_method(old_method, force=True)
        self.assertGreaterEqual(count, 1) 

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def process_images_in_process_pool(images_to_check, number_of_cpus, model):
    if number_of_cpus == -1:
        processes = None
    else:
        processes = number_of_cpus

    # macOS will crash due to a bug in libdispatch if you don't use 'forkserver'
    context = multiprocessing
    if "forkserver" in multiprocessing.get_all_start_methods():
        context = multiprocessing.get_context("forkserver")

    pool = context.Pool(processes=processes)

    function_parameters = zip(
        images_to_check,
        itertools.repeat(model),
    )

    pool.starmap(test_image, function_parameters) 

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def process_images_in_process_pool(images_to_check, known_names, known_face_encodings, number_of_cpus, tolerance, show_distance):
    if number_of_cpus == -1:
        processes = None
    else:
        processes = number_of_cpus

    # macOS will crash due to a bug in libdispatch if you don't use 'forkserver'
    context = multiprocessing
    if "forkserver" in multiprocessing.get_all_start_methods():
        context = multiprocessing.get_context("forkserver")

    pool = context.Pool(processes=processes)

    function_parameters = zip(
        images_to_check,
        itertools.repeat(known_names),
        itertools.repeat(known_face_encodings),
        itertools.repeat(tolerance),
        itertools.repeat(show_distance)
    )

    pool.starmap(test_image, function_parameters) 

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def test_server_multiproc(mocker, set_timeout, restore_signal, start_method):

    mpctx = mp.get_context(start_method)
    mocker.patch('aiotools.server.mp', mpctx)

    started = mpctx.Value('i', 0)
    terminated = mpctx.Value('i', 0)
    proc_idxs = mpctx.Array('i', 3)

    set_timeout(0.2, interrupt)
    aiotools.start_server(myserver_multiproc, num_workers=3,
                          args=(started, terminated, proc_idxs))

    assert started.value == 3
    assert terminated.value == 3
    assert list(proc_idxs) == [0, 1, 2]
    assert len(mp.active_children()) == 0 

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def test_server_multiproc_custom_stop_signals(
        mocker, set_timeout, restore_signal, start_method):

    mpctx = mp.get_context(start_method)
    mocker.patch('aiotools.server.mp', mpctx)

    started = mpctx.Value('i', 0)
    terminated = mpctx.Value('i', 0)
    received_signals = mpctx.Array('i', 2)
    proc_idxs = mpctx.Array('i', 2)

    set_timeout(0.2, interrupt_usr1)
    aiotools.start_server(myserver_multiproc_custom_stop_signals,
                          num_workers=2,
                          stop_signals={signal.SIGUSR1},
                          args=(started, terminated, received_signals, proc_idxs))

    assert started.value == 2
    assert terminated.value == 2
    assert list(received_signals) == [signal.SIGUSR1, signal.SIGUSR1]
    assert list(proc_idxs) == [0, 1]
    assert len(mpctx.active_children()) == 0 

# 需要導入模塊: import multiprocessing [as 別名]
# 或者: from multiprocessing import get_context [as 別名]
def test_external_multi_process_pool_fails_on_write_enabled_checkout(self, repo, backend):
        from multiprocessing import get_context

        co = repo.checkout(write=True)
        co.add_ndarray_column(name='writtenaset', shape=(20, 20), dtype=np.float32, backend=backend)
        with co.columns['writtenaset'] as d:
            for sIdx in range(20):
                d[sIdx] = np.random.randn(20, 20).astype(np.float32) * 100
        assert d.backend == backend
        co.commit(f'master commit number 1')
        co.close()

        nco = repo.checkout(write=True)
        ds = nco.columns['writtenaset']
        keys = [i for i in range(20)]
        with pytest.raises(PermissionError):
            with get_context().Pool(2) as P:
                cmtData = P.map(ds.get, keys)
        nco.close()