Python torch.multiprocessing.SimpleQueue() Examples
The following are 17
code examples of torch.multiprocessing.SimpleQueue().
You can vote up the ones you like or vote down the ones you don't like,
and go to the original project or source file by following the links above each example.
You may also want to check out all available functions/classes of the module
torch.multiprocessing
, or try the search function
.
.
Example #1
| Source File: neat_es.py From DistributedES with Apache License 2.0 | 6 votes |
|
def __init__(self, config):
self.config = config
self.neat_config = self.load_neat_config()
self.neat_config.pop_size = config.pop_size
self.task_q = mp.SimpleQueue()
self.result_q = mp.SimpleQueue()
self.total_steps = 0
stop = mp.Value('i', False)
stats = SharedStats(config.state_dim)
normalizers = [StaticNormalizer(config.state_dim) for _ in range(config.num_workers)]
for normalizer in normalizers:
normalizer.offline_stats.load(stats)
workers = [Worker(id, normalizers[id], self.task_q, self.result_q, stop,
config, self.neat_config) for id in range(config.num_workers)]
for w in workers: w.start()
self.normalizers = normalizers
self.stats = stats
self.stop = stop
Example #2
| Source File: dataloader.py From weakalign with MIT License | 5 votes |
|
def __init__(self, loader):
self.dataset = loader.dataset
self.collate_fn = loader.collate_fn
self.batch_sampler = loader.batch_sampler
self.num_workers = loader.num_workers
self.pin_memory = loader.pin_memory
self.done_event = threading.Event()
self.sample_iter = iter(self.batch_sampler)
if self.num_workers > 0:
self.index_queue = multiprocessing.SimpleQueue()
self.data_queue = multiprocessing.SimpleQueue()
self.batches_outstanding = 0
self.shutdown = False
self.send_idx = 0
self.rcvd_idx = 0
self.reorder_dict = {}
self.workers = [
multiprocessing.Process(
target=_worker_loop,
args=(self.dataset, self.index_queue, self.data_queue, self.collate_fn, np.random.randint(0, 4294967296, dtype='uint32')))
for _ in range(self.num_workers)]
for w in self.workers:
w.daemon = True # ensure that the worker exits on process exit
w.start()
if self.pin_memory:
in_data = self.data_queue
self.data_queue = queue.Queue()
self.pin_thread = threading.Thread(
target=_pin_memory_loop,
args=(in_data, self.data_queue, self.done_event))
self.pin_thread.daemon = True
self.pin_thread.start()
# prime the prefetch loop
for _ in range(2 * self.num_workers):
self._put_indices()
Example #3
| Source File: pytorch_dataloader.py From dlupi-heteroscedastic-dropout with MIT License | 5 votes |
|
def __init__(self, loader):
self.dataset = loader.dataset
self.collate_fn = loader.collate_fn
self.batch_sampler = loader.batch_sampler
self.num_workers = loader.num_workers
self.pin_memory = loader.pin_memory
self.done_event = threading.Event()
self.sample_iter = iter(self.batch_sampler)
if self.num_workers > 0:
self.index_queue = multiprocessing.SimpleQueue()
self.data_queue = multiprocessing.SimpleQueue()
self.batches_outstanding = 0
self.shutdown = False
self.send_idx = 0
self.rcvd_idx = 0
self.reorder_dict = {}
self.workers = [
multiprocessing.Process(
target=_worker_loop,
args=(self.dataset, self.index_queue, self.data_queue, self.collate_fn))
for _ in range(self.num_workers)]
for w in self.workers:
w.daemon = True # ensure that the worker exits on process exit
w.start()
if self.pin_memory:
in_data = self.data_queue
self.data_queue = queue.Queue()
self.pin_thread = threading.Thread(
target=_pin_memory_loop,
args=(in_data, self.data_queue, self.done_event))
self.pin_thread.daemon = True
self.pin_thread.start()
# prime the prefetch loop
for _ in range(2 * self.num_workers):
self._put_indices()
Example #4
| Source File: my_data_loader.py From ps_pytorch with MIT License | 5 votes |
|
def __init__(self, loader):
self.dataset = loader.dataset
self.collate_fn = loader.collate_fn
self.batch_sampler = loader.batch_sampler
self.num_workers = loader.num_workers
self.pin_memory = loader.pin_memory
self.done_event = threading.Event()
self.sample_iter = iter(self.batch_sampler)
if self.num_workers > 0:
self.index_queue = multiprocessing.SimpleQueue()
self.data_queue = multiprocessing.SimpleQueue()
self.batches_outstanding = 0
self.shutdown = False
self.send_idx = 0
self.rcvd_idx = 0
self.reorder_dict = {}
self.workers = [
multiprocessing.Process(
target=_worker_loop,
args=(self.dataset, self.index_queue, self.data_queue, self.collate_fn))
for _ in range(self.num_workers)]
for w in self.workers:
w.daemon = True # ensure that the worker exits on process exit
w.start()
if self.pin_memory:
in_data = self.data_queue
self.data_queue = queue.Queue()
self.pin_thread = threading.Thread(
target=_pin_memory_loop,
args=(in_data, self.data_queue, self.done_event))
self.pin_thread.daemon = True
self.pin_thread.start()
# prime the prefetch loop
for _ in range(2 * self.num_workers):
self._put_indices()
Example #5
| Source File: dataloader.py From keyphrase-gan with MIT License | 5 votes |
|
def __init__(self, loader):
self.dataset = loader.dataset
self.collate_fn = loader.collate_fn
self.batch_sampler = loader.batch_sampler
self.num_workers = loader.num_workers
self.pin_memory = loader.pin_memory
self.done_event = threading.Event()
self.sample_iter = iter(self.batch_sampler)
if self.num_workers > 0:
self.index_queue = multiprocessing.SimpleQueue()
self.data_queue = multiprocessing.SimpleQueue()
self.batches_outstanding = 0
self.shutdown = False
self.send_idx = 0
self.rcvd_idx = 0
self.reorder_dict = {}
self.workers = [
multiprocessing.Process(
target=_worker_loop,
args=(self.dataset, self.index_queue, self.data_queue, self.collate_fn))
for _ in range(self.num_workers)]
for w in self.workers:
w.daemon = True # ensure that the worker exits on process exit
w.start()
if self.pin_memory:
in_data = self.data_queue
self.data_queue = queue.Queue()
self.pin_thread = threading.Thread(
target=_pin_memory_loop,
args=(in_data, self.data_queue, self.done_event))
self.pin_thread.daemon = True
self.pin_thread.start()
# prime the prefetch loop
for _ in range(2 * self.num_workers):
self._put_indices()
Example #6
| Source File: dataloader.py From keyphrase-generation-rl with MIT License | 5 votes |
|
def __init__(self, loader):
self.dataset = loader.dataset
self.collate_fn = loader.collate_fn
self.batch_sampler = loader.batch_sampler
self.num_workers = loader.num_workers
self.pin_memory = loader.pin_memory
self.done_event = threading.Event()
self.sample_iter = iter(self.batch_sampler)
if self.num_workers > 0:
self.index_queue = multiprocessing.SimpleQueue()
self.data_queue = multiprocessing.SimpleQueue()
self.batches_outstanding = 0
self.shutdown = False
self.send_idx = 0
self.rcvd_idx = 0
self.reorder_dict = {}
self.workers = [
multiprocessing.Process(
target=_worker_loop,
args=(self.dataset, self.index_queue, self.data_queue, self.collate_fn))
for _ in range(self.num_workers)]
for w in self.workers:
w.daemon = True # ensure that the worker exits on process exit
w.start()
if self.pin_memory:
in_data = self.data_queue
self.data_queue = queue.Queue()
self.pin_thread = threading.Thread(
target=_pin_memory_loop,
args=(in_data, self.data_queue, self.done_event))
self.pin_thread.daemon = True
self.pin_thread.start()
# prime the prefetch loop
for _ in range(2 * self.num_workers):
self._put_indices()
Example #7
| Source File: dataloader.py From TAKG with MIT License | 5 votes |
|
def __init__(self, loader):
self.dataset = loader.dataset
self.collate_fn = loader.collate_fn
self.batch_sampler = loader.batch_sampler
self.num_workers = loader.num_workers
self.pin_memory = loader.pin_memory
self.done_event = threading.Event()
self.sample_iter = iter(self.batch_sampler)
if self.num_workers > 0:
self.index_queue = multiprocessing.SimpleQueue()
self.data_queue = multiprocessing.SimpleQueue()
self.batches_outstanding = 0
self.shutdown = False
self.send_idx = 0
self.rcvd_idx = 0
self.reorder_dict = {}
self.workers = [
multiprocessing.Process(
target=_worker_loop,
args=(self.dataset, self.index_queue, self.data_queue, self.collate_fn))
for _ in range(self.num_workers)]
for w in self.workers:
w.daemon = True # ensure that the worker exits on process exit
w.start()
if self.pin_memory:
in_data = self.data_queue
self.data_queue = queue.Queue()
self.pin_thread = threading.Thread(
target=_pin_memory_loop,
args=(in_data, self.data_queue, self.done_event))
self.pin_thread.daemon = True
self.pin_thread.start()
# prime the prefetch loop
for _ in range(2 * self.num_workers):
self._put_indices()
Example #8
| Source File: metrics.py From ParlAI with MIT License | 5 votes |
|
def __init__(self, threadsafe=False, shared=None):
self._threadsafe = threadsafe
if self._threadsafe and shared is None:
# Threadsafe metrics tracking works by keeping a queue that workers can
# push updates to. the main worker works through the queue at report
# time. We could add some buffering to improve performance, but we
# are deprioritizing hogwild performance at this time.
self._buffer = None
self._queue = multiprocessing.SimpleQueue()
self._worker = False
self._data = {}
elif shared and 'queue' in shared:
# This is a clone, in threadsafe mode
self._buffer = {}
self._queue = shared['queue']
self._worker = True
self._data = None
elif shared and 'data' in shared:
# This is a clone, in non-threadsafe mode
self._buffer = None
self._queue = None
self._worker = False
self._data = shared['data']
else:
# The original in non-threadsafe mode
self._buffer = None
self._queue = None
self._worker = False
self._data = {}
Example #9
| Source File: dataloader_new.py From MetaFGNet with MIT License | 5 votes |
|
def __init__(self, loader):
self.dataset = loader.dataset
self.collate_fn = loader.collate_fn
self.batch_sampler = loader.batch_sampler
self.num_workers = loader.num_workers
self.pin_memory = loader.pin_memory
self.done_event = threading.Event()
self.sample_iter = iter(self.batch_sampler)
if self.num_workers > 0:
self.index_queue = multiprocessing.SimpleQueue()
self.data_queue = multiprocessing.SimpleQueue()
self.batches_outstanding = 0
self.shutdown = False
self.send_idx = 0
self.rcvd_idx = 0
self.reorder_dict = {}
self.workers = [
multiprocessing.Process(
target=_worker_loop,
args=(self.dataset, self.index_queue, self.data_queue, self.collate_fn))
for _ in range(self.num_workers)]
for w in self.workers:
w.daemon = True # ensure that the worker exits on process exit
w.start()
if self.pin_memory:
in_data = self.data_queue
self.data_queue = queue.Queue()
self.pin_thread = threading.Thread(
target=_pin_memory_loop,
args=(in_data, self.data_queue, self.done_event))
self.pin_thread.daemon = True
self.pin_thread.start()
# prime the prefetch loop
for _ in range(2 * self.num_workers):
self._put_indices()
Example #10
| Source File: dataloader.py From seq2seq-keyphrase-pytorch with Apache License 2.0 | 5 votes |
|
def __init__(self, loader):
self.dataset = loader.dataset
self.collate_fn = loader.collate_fn
self.batch_sampler = loader.batch_sampler
self.num_workers = loader.num_workers
self.pin_memory = loader.pin_memory
self.done_event = threading.Event()
self.sample_iter = iter(self.batch_sampler)
if self.num_workers > 0:
self.index_queue = multiprocessing.SimpleQueue()
self.data_queue = multiprocessing.SimpleQueue()
self.batches_outstanding = 0
self.shutdown = False
self.send_idx = 0
self.rcvd_idx = 0
self.reorder_dict = {}
self.workers = [
multiprocessing.Process(
target=_worker_loop,
args=(self.dataset, self.index_queue, self.data_queue, self.collate_fn))
for _ in range(self.num_workers)]
for w in self.workers:
w.daemon = True # ensure that the worker exits on process exit
w.start()
if self.pin_memory:
in_data = self.data_queue
self.data_queue = queue.Queue()
self.pin_thread = threading.Thread(
target=_pin_memory_loop,
args=(in_data, self.data_queue, self.done_event))
self.pin_thread.daemon = True
self.pin_thread.start()
# prime the prefetch loop
for _ in range(2 * self.num_workers):
self._put_indices()
Example #11
| Source File: monobeast.py From torchbeast with Apache License 2.0 | 5 votes |
|
def get_batch(
flags,
free_queue: mp.SimpleQueue,
full_queue: mp.SimpleQueue,
buffers: Buffers,
initial_agent_state_buffers,
timings,
lock=threading.Lock(),
):
with lock:
timings.time("lock")
indices = [full_queue.get() for _ in range(flags.batch_size)]
timings.time("dequeue")
batch = {
key: torch.stack([buffers[key][m] for m in indices], dim=1) for key in buffers
}
initial_agent_state = (
torch.cat(ts, dim=1)
for ts in zip(*[initial_agent_state_buffers[m] for m in indices])
)
timings.time("batch")
for m in indices:
free_queue.put(m)
timings.time("enqueue")
batch = {k: t.to(device=flags.device, non_blocking=True) for k, t in batch.items()}
initial_agent_state = tuple(
t.to(device=flags.device, non_blocking=True) for t in initial_agent_state
)
timings.time("device")
return batch, initial_agent_state
Example #12
| Source File: dataloader.py From Dense-CoAttention-Network with MIT License | 4 votes |
|
def __init__(self, loader):
self.dataset = loader.dataset
self.collate_fn = loader.collate_fn
self.batch_sampler = loader.batch_sampler
self.num_workers = loader.num_workers
self.pin_memory = loader.pin_memory
self.use_thread = loader.use_thread
self.pin_done_event = threading.Event()
self.sample_iter = iter(self.batch_sampler)
if self.num_workers > 0:
self.index_queue = queue.Queue() if self.use_thread else multiprocessing.SimpleQueue()
self.data_queue = queue.Queue() if self.use_thread else multiprocessing.SimpleQueue()
self.batches_outstanding = 0
self.shutdown = False
self.send_idx = 0
self.rcvd_idx = 0
self.reorder_dict = {}
if self.use_thread:
self.workers = [
threading.Thread(
target=_worker_thread_loop,
args=(self.dataset, self.index_queue, self.data_queue, self.collate_fn))
for _ in range(self.num_workers)]
else:
self.workers = [
multiprocessing.Process(
target=_worker_process_loop,
args=(self.dataset, self.index_queue, self.data_queue, self.collate_fn))
for _ in range(self.num_workers)]
for w in self.workers:
w.daemon = True
w.start()
if self.pin_memory:
in_data = self.data_queue
self.data_queue = queue.Queue()
self.pin_thread = threading.Thread(
target=_pin_memory_loop,
args=(in_data, self.data_queue, self.pin_done_event))
self.pin_thread.daemon = True
self.pin_thread.start()
for _ in range(2 * self.num_workers):
self._put_indices()
Example #13
| Source File: dataloader.py From semantic-segmentation-pytorch with BSD 3-Clause "New" or "Revised" License | 4 votes |
|
def __init__(self, loader):
self.dataset = loader.dataset
self.collate_fn = loader.collate_fn
self.batch_sampler = loader.batch_sampler
self.num_workers = loader.num_workers
self.pin_memory = loader.pin_memory and torch.cuda.is_available()
self.timeout = loader.timeout
self.done_event = threading.Event()
self.sample_iter = iter(self.batch_sampler)
if self.num_workers > 0:
self.worker_init_fn = loader.worker_init_fn
self.index_queue = multiprocessing.SimpleQueue()
self.worker_result_queue = multiprocessing.SimpleQueue()
self.batches_outstanding = 0
self.worker_pids_set = False
self.shutdown = False
self.send_idx = 0
self.rcvd_idx = 0
self.reorder_dict = {}
base_seed = torch.LongTensor(1).random_(0, 2**31-1)[0]
self.workers = [
multiprocessing.Process(
target=_worker_loop,
args=(self.dataset, self.index_queue, self.worker_result_queue, self.collate_fn,
base_seed + i, self.worker_init_fn, i))
for i in range(self.num_workers)]
if self.pin_memory or self.timeout > 0:
self.data_queue = queue.Queue()
if self.pin_memory:
maybe_device_id = torch.cuda.current_device()
else:
# do not initialize cuda context if not necessary
maybe_device_id = None
self.worker_manager_thread = threading.Thread(
target=_worker_manager_loop,
args=(self.worker_result_queue, self.data_queue, self.done_event, self.pin_memory,
maybe_device_id))
self.worker_manager_thread.daemon = True
self.worker_manager_thread.start()
else:
self.data_queue = self.worker_result_queue
for w in self.workers:
w.daemon = True # ensure that the worker exits on process exit
w.start()
_set_worker_pids(id(self), tuple(w.pid for w in self.workers))
_set_SIGCHLD_handler()
self.worker_pids_set = True
# prime the prefetch loop
for _ in range(2 * self.num_workers):
self._put_indices()
Example #14
| Source File: loader_fn.py From 3D-convolutional-speaker-recognition-pytorch with Apache License 2.0 | 4 votes |
|
def __init__(self, loader):
self.dataset = loader.dataset
self.collate_fn = loader.collate_fn
self.batch_sampler = loader.batch_sampler
self.num_workers = loader.num_workers
self.pin_memory = loader.pin_memory and torch.cuda.is_available()
self.timeout = loader.timeout
self.done_event = threading.Event()
self.sample_iter = iter(self.batch_sampler)
if self.num_workers > 0:
self.worker_init_fn = loader.worker_init_fn
self.index_queues = [multiprocessing.SimpleQueue() for _ in range(self.num_workers)]
self.worker_queue_idx = 0
self.worker_result_queue = multiprocessing.SimpleQueue()
self.batches_outstanding = 0
self.worker_pids_set = False
self.shutdown = False
self.send_idx = 0
self.rcvd_idx = 0
self.reorder_dict = {}
base_seed = torch.LongTensor(1).random_()[0]
self.workers = [
multiprocessing.Process(
target=_worker_loop,
args=(self.dataset, self.index_queues[i],
self.worker_result_queue, self.collate_fn, base_seed + i,
self.worker_init_fn, i))
for i in range(self.num_workers)]
if self.pin_memory or self.timeout > 0:
self.data_queue = queue.Queue()
if self.pin_memory:
maybe_device_id = torch.cuda.current_device()
else:
# do not initialize cuda context if not necessary
maybe_device_id = None
self.worker_manager_thread = threading.Thread(
target=_worker_manager_loop,
args=(self.worker_result_queue, self.data_queue, self.done_event, self.pin_memory,
maybe_device_id))
self.worker_manager_thread.daemon = True
self.worker_manager_thread.start()
else:
self.data_queue = self.worker_result_queue
for w in self.workers:
w.daemon = True # ensure that the worker exits on process exit
w.start()
_update_worker_pids(id(self), tuple(w.pid for w in self.workers))
_set_SIGCHLD_handler()
self.worker_pids_set = True
# prime the prefetch loop
for _ in range(2 * self.num_workers):
self._put_indices()
Example #15
| Source File: dataloader.py From EMANet with GNU General Public License v3.0 | 4 votes |
|
def __init__(self, loader):
self.dataset = loader.dataset
self.collate_fn = loader.collate_fn
self.batch_sampler = loader.batch_sampler
self.num_workers = loader.num_workers
self.pin_memory = loader.pin_memory and torch.cuda.is_available()
self.timeout = loader.timeout
self.done_event = threading.Event()
self.sample_iter = iter(self.batch_sampler)
if self.num_workers > 0:
self.worker_init_fn = loader.worker_init_fn
self.index_queue = multiprocessing.SimpleQueue()
self.worker_result_queue = multiprocessing.SimpleQueue()
self.batches_outstanding = 0
self.worker_pids_set = False
self.shutdown = False
self.send_idx = 0
self.rcvd_idx = 0
self.reorder_dict = {}
base_seed = torch.LongTensor(1).random_(0, 2**31-1)[0]
self.workers = [
multiprocessing.Process(
target=_worker_loop,
args=(self.dataset, self.index_queue, self.worker_result_queue, self.collate_fn,
base_seed + i, self.worker_init_fn, i))
for i in range(self.num_workers)]
if self.pin_memory or self.timeout > 0:
self.data_queue = queue.Queue()
if self.pin_memory:
maybe_device_id = torch.cuda.current_device()
else:
# do not initialize cuda context if not necessary
maybe_device_id = None
self.worker_manager_thread = threading.Thread(
target=_worker_manager_loop,
args=(self.worker_result_queue, self.data_queue, self.done_event, self.pin_memory,
maybe_device_id))
self.worker_manager_thread.daemon = True
self.worker_manager_thread.start()
else:
self.data_queue = self.worker_result_queue
for w in self.workers:
w.daemon = True # ensure that the worker exits on process exit
w.start()
_update_worker_pids(id(self), tuple(w.pid for w in self.workers))
_set_SIGCHLD_handler()
self.worker_pids_set = True
# prime the prefetch loop
for _ in range(2 * self.num_workers):
self._put_indices()
Example #16
| Source File: dataloader.py From mt-dnn with MIT License | 4 votes |
|
def __init__(self, loader):
self.dataset = loader.dataset
self.collate_fn = loader.collate_fn
self.batch_sampler = loader.batch_sampler
self.num_workers = loader.num_workers
self.pin_memory = loader.pin_memory and torch.cuda.is_available()
self.timeout = loader.timeout
self.done_event = threading.Event()
self.sample_iter = iter(self.batch_sampler)
if self.num_workers > 0:
self.worker_init_fn = loader.worker_init_fn
self.index_queues = [multiprocessing.Queue() for _ in range(self.num_workers)]
self.worker_queue_idx = 0
self.worker_result_queue = multiprocessing.SimpleQueue()
self.batches_outstanding = 0
self.worker_pids_set = False
self.shutdown = False
self.send_idx = 0
self.rcvd_idx = 0
self.reorder_dict = {}
self.workers = [
multiprocessing.Process(
target=_worker_loop,
args=(self.dataset, self.index_queues[i],
self.worker_result_queue, self.collate_fn, self.worker_init_fn, i))
for i in range(self.num_workers)]
if self.pin_memory or self.timeout > 0:
self.data_queue = queue.Queue()
if self.pin_memory:
maybe_device_id = torch.cuda.current_device()
else:
# do not initialize cuda context if not necessary
maybe_device_id = None
self.worker_manager_thread = threading.Thread(
target=_worker_manager_loop,
args=(self.worker_result_queue, self.data_queue, self.done_event, self.pin_memory,
maybe_device_id))
self.worker_manager_thread.daemon = True
self.worker_manager_thread.start()
else:
self.data_queue = self.worker_result_queue
for w in self.workers:
w.daemon = True # ensure that the worker exits on process exit
w.start()
_set_worker_pids(id(self), tuple(w.pid for w in self.workers))
_set_SIGCHLD_handler()
self.worker_pids_set = True
# prime the prefetch loop
for _ in range(2 * self.num_workers):
self._put_indices()
Example #17
| Source File: multiproc.py From proSR with GNU General Public License v3.0 | 4 votes |
|
def __init__(self, loader):
self.random_vars = loader.random_vars
self.dataset = loader.dataset
self.collate_fn = loader.collate_fn
self.batch_sampler = loader.batch_sampler
self.num_workers = loader.num_workers
self.pin_memory = loader.pin_memory and torch.cuda.is_available()
self.timeout = loader.timeout
self.done_event = threading.Event()
self.sample_iter = iter(self.batch_sampler)
base_seed = torch.LongTensor(1).random_().item()
if self.num_workers > 0:
self.worker_init_fn = loader.worker_init_fn
self.index_queues = [multiprocessing.Queue() for _ in range(self.num_workers)]
self.worker_queue_idx = 0
self.worker_result_queue = multiprocessing.SimpleQueue()
self.batches_outstanding = 0
self.worker_pids_set = False
self.shutdown = False
self.send_idx = 0
self.rcvd_idx = 0
self.reorder_dict = {}
self.workers = [
multiprocessing.Process(
target=_worker_loop,
args=(self.dataset, self.index_queues[i],
self.worker_result_queue, self.collate_fn, base_seed + i,
self.worker_init_fn, i))
for i in range(self.num_workers)]
if self.pin_memory or self.timeout > 0:
self.data_queue = queue.Queue()
if self.pin_memory:
maybe_device_id = torch.cuda.current_device()
else:
# do not initialize cuda context if not necessary
maybe_device_id = None
self.worker_manager_thread = threading.Thread(
target=_worker_manager_loop,
args=(self.worker_result_queue, self.data_queue, self.done_event, self.pin_memory,
maybe_device_id))
self.worker_manager_thread.daemon = True
self.worker_manager_thread.start()
else:
self.data_queue = self.worker_result_queue
for w in self.workers:
w.daemon = True # ensure that the worker exits on process exit
w.start()
_update_worker_pids(id(self), tuple(w.pid for w in self.workers))
_set_SIGCHLD_handler()
self.worker_pids_set = True
# prime the prefetch loop
for _ in range(2 * self.num_workers):
self._put_indices()
