import collections
import contextlib
import warnings
import ctypes
import pickle
import sys
import os
import stat
import platform
from typing import Any, Dict, Optional, Tuple, Union
import torch_npu
from torch_npu.utils._error_code import ErrCode, pta_error
from . import is_initialized, _get_device_index, _lazy_init
from .utils import _dummy_type
from ._memory_viz import memory as _memory, segments as _segments
__all__ = [
"caching_allocator_alloc",
"caching_allocator_delete",
"get_per_process_memory_fraction",
"set_per_process_memory_fraction",
"empty_cache",
"empty_virt_addr_cache",
"memory_stats",
"memory_stats_as_nested_dict",
"reset_accumulated_memory_stats",
"reset_peak_memory_stats",
"reset_max_memory_allocated",
"reset_max_memory_cached",
"memory_allocated",
"max_memory_allocated",
"memory_reserved",
"max_memory_reserved",
"memory_cached",
"max_memory_cached",
"memory_snapshot",
"memory_summary",
"get_allocator_backend",
"NPUPluggableAllocator",
"change_current_allocator",
"MemPool",
"MemPoolContext",
"use_mem_pool",
"host_empty_cache",
"host_memory_stats",
"host_memory_stats_as_nested_dict",
"reset_accumulated_host_memory_stats",
"reset_peak_host_memory_stats"
]
if not hasattr(torch_npu._C, "_npu_NPUAllocator"):
torch_npu._C.__dict__["_npu_NPUAllocator"] = _dummy_type("_npu_NPUAllocator")
if not hasattr(torch_npu._C, "_MemPool"):
torch_npu._C.__dict__["_MemPool"] = _dummy_type("_MemPool")
torch_npu._C.__dict__["_MemPoolContext"] = _dummy_type("_MemPoolContext")
torch_npu._C.__dict__["_npu_beginAllocateToPool"] = _dummy_type(
"_npu_beginAllocateToPool"
)
torch_npu._C.__dict__["_npu_endAllocateCurrentStreamToPool"] = _dummy_type(
"_npu_endAllocateCurrentStreamToPool"
)
@contextlib.contextmanager
def _free_mutex():
torch_npu._C._npu_lock_mutex()
try:
yield
finally:
torch_npu._C._npu_unlock_mutex()
def caching_allocator_alloc(size, device=None, stream=None):
r"""Performs a memory allocation using the NPU memory allocator.
Memory is allocated for a given device and a stream, this
function is intended to be used for interoperability with other
frameworks. Allocated memory is released through
:func:`~torch_npu.npu.caching_allocator_delete`.
Arguments:
size (int): number of bytes to be allocated.
device (torch.device or int, optional): selected device. If it is
``None`` the default NPU device is used.
stream (torch_npu.npu.Stream or int, optional): selected stream. If is ``None`` then
the default stream for the selected device is used.
.. note::
See :ref:`npu-memory-management` for more details about NPU memory
management.
"""
if device is None:
device = torch_npu.npu.current_device()
device = _get_device_index(device)
if stream is None:
stream = torch_npu.npu.current_stream(device)
if isinstance(stream, torch_npu.npu.streams.Stream):
stream = stream.npu_stream
if not isinstance(stream, int):
raise TypeError('Invalid type for stream argument, must be '
'`torch_npu.npu.Stream` or `int` representing a pointer '
'to a exisiting stream' + pta_error(ErrCode.TYPE))
with torch_npu.npu.device(device):
return torch_npu._C._npu_npuCachingAllocator_raw_alloc(size, stream)
def caching_allocator_delete(mem_ptr):
r"""Deletes memory allocated using the NPU memory allocator.
Memory allocated with :func:`~torch_npu.npu.caching_allocator_alloc`.
is freed here. The associated device and stream are tracked inside
the allocator.
Arguments:
mem_ptr (int): memory address to be freed by the allocator.
.. note::
See :ref:`npu-memory-management` for more details about NPU memory
management.
"""
torch_npu._C._npu_npuCachingAllocator_raw_delete(mem_ptr)
def get_per_process_memory_fraction(device=None) -> float:
r"""Get memory fraction for a process.
Args:
device (torch.device or int, optional): selected device. If it is
``None`` the default NPU device is used.
Returns:
memory fraction, in range 0~1. Allowed memory equals total_memory * fraction.
"""
_lazy_init()
if device is None:
device = torch_npu.npu.current_device()
device = _get_device_index(device)
return torch_npu._C._npu_getMemoryFraction(device)
def set_per_process_memory_fraction(fraction, device=None) -> None:
r"""Set memory fraction for a process.
The fraction is used to limit an caching allocator to allocated memory on a NPU device.
The allowed value equals the total visible memory multiplied fraction.
If trying to allocate more than the allowed value in a process, will raise an out of
memory error in allocator.
Arguments:
fraction(float): Range: 0~1. Allowed memory equals total_memory * fraction.
device (torch.device or int, optional): selected device. If it is
``None`` the default NPU device is used.
.. note::
In general, the total available free memory is less than the total capacity.
"""
_lazy_init()
if device is None:
device = torch_npu.npu.current_device()
device = _get_device_index(device)
if not isinstance(fraction, float):
raise TypeError('Invalid type for fraction argument, must be `float`' + pta_error(ErrCode.TYPE))
if fraction < 0 or fraction > 1:
raise ValueError('Invalid fraction value: {}. '
'Allowed range: 0~1'.format(fraction) + pta_error(ErrCode.VALUE))
torch_npu._C._npu_setMemoryFraction(fraction, device)
def empty_cache():
r"""Releases all unoccupied cached memory currently held by the caching
allocator so that those can be used in other NPU application and visible in
`nvidia-smi`.
.. note::
:func:`~torch_npu.npu.empty_cache` doesn't increase the amount of NPU
memory available for PyTorch. However, it may help reduce fragmentation
of NPU memory in certain cases. See :ref:`npu-memory-management` for
more details about NPU memory management.
"""
if is_initialized():
torch_npu._C._npu_emptyCache()
def host_empty_cache():
r"""
Releases all unoccupied cached memory currently held by the caching
allocator so that those can be used in other NPU application.
"""
torch_npu._C._npu_hostEmptyCache()
def host_memory_stats():
r"""Return a dictionary of pinned (host) allocator statistics.
Core statistics (host pinned allocator):
- ``"allocations.{current,peak,allocated,freed}"``:
pinned blocks owned by the allocator (active + cached). Grows when a new
block is created via NPU and shrinks when cached blocks are returned.
- ``"allocated_bytes.{current,peak,allocated,freed}"``:
bytes of pinned blocks owned by the allocator (active + cached), using
the rounded block size requested from NPU.
- ``"active_requests.{current,peak,allocated,freed}"``:
blocks currently checked out to callers (increments on handout, decrements
when the block becomes reusable after stream deps finish).
- ``"active_bytes.{current,peak,allocated,freed}"``:
bytes corresponding to active blocks.
Metric type:
- ``current``: current value.
- ``peak``: maximum value.
- ``allocated``: historical total increase.
- ``freed``: historical total decrease.
Event/timing counters:
- ``"num_host_alloc"`` / ``"num_host_free"``: blocks created to grow the
pool / cached blocks returned to NPU (matches allocations allocated/freed).
- ``"host_alloc_time.{total,max,min,count,avg}"``: time in host alloc calls
when growing the pool (microseconds).
- ``"host_free_time.{total,max,min,count,avg}"``: time in host free calls
when cached blocks are returned (microseconds).
Block sizes are rounded up to the next power of two before calling NPU, so
byte stats reflect the rounded size. Peak values are aggregated per bucket
and are a best-effort approximation of the true peak.
"""
result = []
def _recurse_add_to_result(prefix, obj):
if isinstance(obj, dict):
if len(prefix) > 0:
prefix += "."
for k, v in obj.items():
_recurse_add_to_result(prefix + k, v)
else:
result.append((prefix, obj))
stats = host_memory_stats_as_nested_dict()
_recurse_add_to_result("", stats)
result.sort()
return collections.OrderedDict(result)
def host_memory_stats_as_nested_dict():
r"""Return the result of :func:`~torch_npu.npu.host_memory_stats` as a nested dictionary."""
return torch_npu._C._npu_hostMemoryStats()
def reset_accumulated_host_memory_stats():
r"""Reset the "accumulated" (historical) stats tracked by the host memory allocator.
See :func:`~torch_npu.npu.host_memory_stats` for details. Accumulated stats correspond to
the `"allocated"` and `"freed"` keys in each individual stat dict.
"""
return torch_npu._C._npu_resetAccumulatedHostMemoryStats()
def reset_peak_host_memory_stats():
r"""Reset the "peak" stats tracked by the host memory allocator.
See :func:`~torch_npu.npu.host_memory_stats` for details. Peak stats correspond to the
`"peak"` key in each individual stat dict.
"""
return torch_npu._C._npu_resetPeakHostMemoryStats()
def empty_virt_addr_cache():
r"""Light-weight version of empty_cache(). It only unmaps virtual address,
and store the free physical handles for later malloc.
"""
if is_initialized():
torch_npu._C._npu_emptyVirtAddrCache()
def memory_stats(device=None):
"""Returns a dictionary of NPU memory allocator statistics for a
given device.
The return value of this function is a dictionary of statistics, each of
which is a non-negative integer.
Core statistics:
- ``"allocated.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:
number of allocation requests received by the memory allocator.
- ``"allocated_bytes.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:
amount of allocated memory.
- ``"segment.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:
number of reserved segments from ``npuMalloc()``.
- ``"reserved_bytes.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:
amount of reserved memory.
- ``"active.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:
number of active memory blocks.
- ``"active_bytes.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:
amount of active memory.
- ``"inactive_split.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:
number of inactive, non-releasable memory blocks.
- ``"inactive_split_bytes.{all,large_pool,small_pool}.{current,peak,allocated,freed}"``:
amount of inactive, non-releasable memory.
For these core statistics, values are broken down as follows.
Pool type:
- ``all``: combined statistics across all memory pools.
- ``large_pool``: statistics for the large allocation pool
(as of October 2019, for size >= 1MB allocations).
- ``small_pool``: statistics for the small allocation pool
(as of October 2019, for size < 1MB allocations).
Metric type:
- ``current``: current value of this metric.
- ``peak``: maximum value of this metric.
- ``allocated``: historical total increase in this metric.
- ``freed``: historical total decrease in this metric.
In addition to the core statistics, we also provide some simple event
counters:
- ``"num_alloc_retries"``: number of failed ``npuMalloc`` calls that
result in a cache flush and retry.
- ``"num_ooms"``: number of out-of-memory errors thrown.
The caching allocator can be configured via ENV to not split blocks larger than a
defined size (see Memory Management section of the Cuda Semantics documentation).
This helps avoid memory framentation but may have a performance
penalty. Additional outputs to assist with tuning and evaluating impact:
- ``"max_split_size"``: blocks above this size will not be split.
- ``"oversize_allocations.{current,peak,allocated,freed}"``:
number of over-size allocation requests received by the memory allocator.
- ``"oversize_segments.{current,peak,allocated,freed}"``:
number of over-size reserved segments from ``cudaMalloc()``.
Arguments:
device (torch.device or int, optional): selected device. Returns
statistics for the current device, given by :func:`~torch_npu.npu.current_device`,
if :attr:`device` is ``None`` (default).
.. note::
See :ref:`npu-memory-management` for more details about NPU memory
management.
"""
result = []
def _recurse_add_to_result(prefix, obj):
if isinstance(obj, dict):
if len(prefix) > 0:
prefix += "."
for k, v in obj.items():
_recurse_add_to_result(prefix + k, v)
else:
result.append((prefix, obj))
stats = memory_stats_as_nested_dict(device=device)
_recurse_add_to_result("", stats)
result.sort()
return collections.OrderedDict(result)
def memory_stats_as_nested_dict(device=None):
r"""Returns the result of :func:`~torch_npu.npu.memory_stats` as a nested dictionary."""
if not is_initialized():
return {}
device = _get_device_index(device, optional=True)
return torch_npu._C._npu_memoryStats(device)
def reset_accumulated_memory_stats(device=None):
r"""Resets the "accumulated" (historical) stats tracked by the NPU memory allocator.
See :func:`~torch_npu.npu.memory_stats` for details. Accumulated stats correspond to
the `"allocated"` and `"freed"` keys in each individual stat dict, as well as
`"num_alloc_retries"` and `"num_ooms"`.
Arguments:
device (torch.device or int, optional): selected device. Returns
statistic for the current device, given by :func:`~torch_npu.npu.current_device`,
if :attr:`device` is ``None`` (default).
.. note::
See :ref:`npu-memory-management` for more details about NPU memory
management.
"""
device = _get_device_index(device, optional=True)
return torch_npu._C._npu_resetAccumulatedMemoryStats(device)
def reset_peak_memory_stats(device=None):
r"""Resets the "peak" stats tracked by the NPU memory allocator.
See :func:`~torch_npu.npu.memory_stats` for details. Peak stats correspond to the
`"peak"` key in each individual stat dict.
Arguments:
device (torch.device or int, optional): selected device. Returns
statistic for the current device, given by :func:`~torch_npu.npu.current_device`,
if :attr:`device` is ``None`` (default).
.. note::
See :ref:`npu-memory-management` for more details about NPU memory
management.
"""
device = _get_device_index(device, optional=True)
return torch_npu._C._npu_resetPeakMemoryStats(device)
def reset_max_memory_allocated(device=None):
r"""Resets the starting point in tracking maximum NPU memory occupied by
tensors for a given device.
See :func:`~torch_npu.npu.max_memory_allocated` for details.
Arguments:
device (torch.device or int, optional): selected device. Returns
statistic for the current device, given by :func:`~torch_npu.npu.current_device`,
if :attr:`device` is ``None`` (default).
.. warning::
This function now calls :func:`~torch_npu.npu.reset_peak_memory_stats`, which resets
/all/ peak memory stats.
.. note::
See :ref:`npu-memory-management` for more details about NPU memory
management.
"""
warnings.warn(
"torch_npu.npu.reset_max_memory_allocated now calls torch_npu.npu.reset_peak_memory_stats, "
"which resets /all/ peak memory stats.",
DeprecationWarning)
return reset_peak_memory_stats(device=device)
def reset_max_memory_cached(device=None):
r"""Resets the starting point in tracking maximum NPU memory managed by the
caching allocator for a given device.
See :func:`~torch_npu.npu.max_memory_cached` for details.
Arguments:
device (torch.device or int, optional): selected device. Returns
statistic for the current device, given by :func:`~torch_npu.npu.current_device`,
if :attr:`device` is ``None`` (default).
.. warning::
This function now calls :func:`~torch_npu.npu.reset_peak_memory_stats`, which resets
/all/ peak memory stats.
.. note::
See :ref:`npu-memory-management` for more details about NPU memory
management.
"""
warnings.warn(
"torch_npu.npu.reset_max_memory_cached now calls torch_npu.npu.reset_peak_memory_stats, "
"which resets /all/ peak memory stats.",
DeprecationWarning)
return reset_peak_memory_stats(device=device)
def memory_allocated(device=None):
r"""Returns the current NPU memory occupied by tensors in bytes for a given
device.
Arguments:
device (torch.device or int, optional): selected device. Returns
statistic for the current device, given by :func:`~torch_npu.npu.current_device`,
if :attr:`device` is ``None`` (default).
.. note::
This is likely less than the amount shown in `nvidia-smi` since some
unused memory can be held by the caching allocator and some context
needs to be created on NPU. See :ref:`npu-memory-management` for more
details about NPU memory management.
"""
return memory_stats(device=device).get("allocated_bytes.all.current", 0)
def max_memory_allocated(device=None):
r"""Returns the maximum NPU memory occupied by tensors in bytes for a given
device.
By default, this returns the peak allocated memory since the beginning of
this program. :func:`~torch_npu.npu.reset_peak_stats` can be used to
reset the starting point in tracking this metric. For example, these two
functions can measure the peak allocated memory usage of each iteration in a
training loop.
Arguments:
device (torch.device or int, optional): selected device. Returns
statistic for the current device, given by :func:`~torch_npu.npu.current_device`,
if :attr:`device` is ``None`` (default).
.. note::
See :ref:`npu-memory-management` for more details about NPU memory
management.
"""
return memory_stats(device=device).get("allocated_bytes.all.peak", 0)
def memory_reserved(device=None):
r"""Returns the current NPU memory managed by the caching allocator in bytes
for a given device.
Arguments:
device (torch.device or int, optional): selected device. Returns
statistic for the current device, given by :func:`~torch_npu.npu.current_device`,
if :attr:`device` is ``None`` (default).
.. note::
See :ref:`npu-memory-management` for more details about NPU memory
management.
"""
return memory_stats(device=device).get("reserved_bytes.all.current", 0)
def max_memory_reserved(device=None):
r"""Returns the maximum NPU memory managed by the caching allocator in bytes
for a given device.
By default, this returns the peak cached memory since the beginning of this
program. :func:`~torch_npu.npu.reset_peak_stats` can be used to reset
the starting point in tracking this metric. For example, these two functions
can measure the peak cached memory amount of each iteration in a training
loop.
Arguments:
device (torch.device or int, optional): selected device. Returns
statistic for the current device, given by :func:`~torch_npu.npu.current_device`,
if :attr:`device` is ``None`` (default).
.. note::
See :ref:`npu-memory-management` for more details about NPU memory
management.
"""
return memory_stats(device=device).get("reserved_bytes.all.peak", 0)
def memory_cached(device=None):
r"""Deprecated; see :func:`~torch_npu.npu.memory_reserved`."""
warnings.warn(
"torch_npu.npu.memory_cached has been renamed to torch_npu.npu.memory_reserved",
DeprecationWarning)
return memory_reserved(device=device)
def max_memory_cached(device=None):
r"""Deprecated; see :func:`~torch_npu.npu.max_memory_reserved`."""
warnings.warn(
"torch_npu.npu.max_memory_cached has been renamed to torch_npu.npu.max_memory_reserved",
DeprecationWarning)
return max_memory_reserved(device=device)
def memory_snapshot():
r"""Returns a snapshot of the NPU memory allocator state across all devices.
Interpreting the output of this function requires familiarity with the
memory allocator internals.
.. note::
See :ref:`npu-memory-management` for more details about NPU memory
management.
"""
return torch_npu._C._npu_memorySnapshot()["segments"]
def _format_size(sz, pref_sz):
prefixes = ["B ", "KB", "MB", "GB", "TB", "PB"]
prefix = prefixes[0]
for new_prefix in prefixes[1:]:
if pref_sz < 768 * 1024:
break
prefix = new_prefix
sz //= 1024
pref_sz /= 1024
return "{:7d} {}".format(sz, prefix)
def _format_count(cnt, pref_cnt):
prefixes = [" ", "K", "M"]
prefix = prefixes[0]
for new_prefix in prefixes[1:]:
if pref_cnt < 750 * 1000:
break
prefix = new_prefix
cnt //= 1000
pref_cnt /= 1000
return "{:7d} {} ".format(cnt, prefix)
def _create_metrics_to_display():
metrics_to_display = [
("allocated_bytes", "Allocated memory", _format_size),
("active_bytes", "Active memory", _format_size),
("reserved_bytes", "NPU reserved memory", _format_size),
("inactive_split_bytes", "Non-releasable memory", _format_size),
("allocation", "Allocations", _format_count),
("active", "Active allocs", _format_count),
("segment", "NPU reserved segments", _format_count),
("inactive_split", "Non-releasable allocs", _format_count),
]
lines = []
lines.append("=" * 75)
lines.append(" {_:16} PyTorch NPU memory summary, device ID {device:<18d} ")
lines.append("-" * 75)
lines.append(" {_:9} NPU OOMs: {num_ooms:<13d} | {_:6} npuMalloc retries: {num_alloc_retries:<9d} ")
lines.append("=" * 75)
lines.append(" Metric | Cur Usage | Peak Usage | Tot Alloc | Tot Freed ")
return metrics_to_display, lines
def memory_summary(device=None, abbreviated=False):
r"""Returns a human-readable printout of the current memory allocator
statistics for a given device.
This can be useful to display periodically during training, or when
handling out-of-memory exceptions.
Arguments:
device (torch.device or int, optional): selected device. Returns
printout for the current device, given by :func:`~torch_npu.npu.current_device`,
if :attr:`device` is ``None`` (default).
abbreviated (bool, optional): whether to return an abbreviated summary
(default: False).
.. note::
See :ref:`npu-memory-management` for more details about NPU memory
management.
"""
device = _get_device_index(device, optional=True)
stats = memory_stats(device=device)
metrics_to_display, lines = _create_metrics_to_display()
for metric_key, metric_name, formatter in metrics_to_display:
lines.append("-" * 75)
submetrics = [("all", metric_name)]
if not abbreviated:
submetrics.append(("large_pool", " from large pool"))
submetrics.append(("small_pool", " from small pool"))
current_prefval, peak_prefval, allocated_prefval, freed_prefval = None, None, None, None
for submetric_key, submetric_name in submetrics:
prefix = metric_key + "." + submetric_key + "."
current = stats[prefix + "current"]
peak = stats[prefix + "peak"]
allocated = stats[prefix + "allocated"]
freed = stats[prefix + "freed"]
if current_prefval is None:
current_prefval = current
peak_prefval = peak
allocated_prefval = allocated
freed_prefval = freed
lines.append(" {:<21} | {} | {} | {} | {} ".format(
submetric_name,
formatter(current, current_prefval),
formatter(peak, peak_prefval),
formatter(allocated, allocated_prefval),
formatter(freed, freed_prefval)), )
metrics_to_display = [
("oversize_allocations", "Oversize allocations", _format_count),
("oversize_segments", "Oversize NPU segments", _format_count),
]
for metric_key, metric_name, formatter in metrics_to_display:
lines.append("-" * 75)
prefix = metric_key + "."
current = stats[prefix + "current"]
peak = stats[prefix + "peak"]
allocated = stats[prefix + "allocated"]
freed = stats[prefix + "freed"]
lines.append(" {:<21} | {} | {} | {} | {} ".format(
metric_name,
formatter(current, current),
formatter(peak, peak),
formatter(allocated, allocated),
formatter(freed, freed)), )
lines.append("=" * 75)
fmt_dict = {"_": "", "device": device}
for k, v in stats.items():
fmt_dict[k.replace(".", "-")] = v
return "|" + "|\n|".join(lines).format(**fmt_dict) + "|\n"
def get_allocator_backend() -> str:
r"""Returns a string describing the active allocator backend as set by
``PYTORCH_NPU_ALLOC_CONF``. Currently available backends are
``native`` (PyTorch's native caching allocator).
.. note::
See :ref:`npu-memory-management` for details on choosing the allocator backend.
"""
return torch_npu._C._npu_getAllocatorBackend()
def _set_allocator_settings(settings) -> None:
r"""Sets the allocator settings. Only support expandable_segments:True or False.
"""
return torch_npu._C._npu_npuCachingAllocator_set_allocator_settings(settings)
class _NPUAllocator:
r"""Wrapper over internal NPU memory allocators."""
def __init__(self, allocator: torch_npu._C._npu_NPUAllocator):
self._allocator = allocator
def allocator(self):
return self._allocator
class NPUPluggableAllocator(_NPUAllocator):
r"""NPU memory allocator loaded from a so file."""
def __init__(self, path_to_so_file: str, alloc_fn_name: str, free_fn_name: str):
r"""Memory allocators are compiled in .so files and loaded dynamically using ctypes.
To change the active allocator use the :func:`torch_npu.npu.change_current_allocator` function.
Args:
path_to_so_file(str): Path in the filesystem to the `.so` file containing
the allocator functions
alloc_fn_name(str): Name of the function to perform the memory allocation
in the so file. The signature must be:
void* alloc_fn_name(ssize_t size, int device, aclrtStream stream);
free_fn_name(str): Name of the function to perform the memory release
in the so file. The signature must be:
void free_fn_name(void* ptr, size_t size, aclrtStream stream);
.. warning::
This is currently supported only in unix OSs
.. note::
See :ref:`npu-memory-management` for details on creating and using a custom allocator
"""
allocator = ctypes.CDLL(path_to_so_file)
alloc_fn = ctypes.cast(getattr(allocator, alloc_fn_name), ctypes.c_void_p).value
free_fn = ctypes.cast(getattr(allocator, free_fn_name), ctypes.c_void_p).value
if alloc_fn is None:
raise RuntimeError('alloc_fn is None' + pta_error(ErrCode.NOT_FOUND))
if free_fn is None:
raise RuntimeError('free_fn is None' + pta_error(ErrCode.NOT_FOUND))
self._allocator = torch_npu._C._npu_customAllocator(alloc_fn, free_fn)
def change_current_allocator(allocator: _NPUAllocator) -> None:
r"""Change the currently used memory allocator to be the one provided.
If the current allocator has already been used/initialized, this function will error.
Args:
allocator (torch_npu.npu.memory._NPUAllocator): allocator to be set as the active one.
.. note::
See :ref:`npu-memory-management` for details on creating and using a custom allocator
"""
torch_npu._C._npu_changeCurrentAllocator(allocator.allocator())
def _get_current_allocator() -> _NPUAllocator:
r"""Return the allocator being currently used.
.. note::
See :ref:`npu-memory-management` for details on creating and using a custom allocator
"""
return _NPUAllocator(torch_npu._C._npu_getAllocator())
class MemPool(torch_npu._C._MemPool):
r"""MemPool represents a pool of memory in a caching allocator. Currently,
it's just the ID of the pool object maintained in the NPUCachingAllocator.
Args:
allocator(torch_npu._C._npu_NPUAllocator, optional): a
torch_npu._C._npu_NPUAllocator object that can be used to
define how memory gets allocated in the pool. If :attr:`allocator`
is ``None`` (default), memory allocation follows the default/
current configuration of the NPUCachingAllocator.
"""
def __init__(self, allocator: Optional[torch_npu._C._npu_NPUAllocator] = None):
super().__init__(allocator, True)
@property
def id(self) -> Tuple[int, int]:
r"""Returns the ID of this pool as a tuple of two ints."""
return super().id
@property
def allocator(self) -> Optional[torch_npu._C._npu_NPUAllocator]:
r"""Returns the allocator this MemPool routes allocations to"""
return super().allocator
class MemPoolContext(torch_npu._C._MemPoolContext):
r"""MemPoolContext holds the currently active pool and stashes the previous
pool. On deletion it makes the previous pool active.
Args:
pool(torch_npu.npu.MemPool): a MemPool object to be made active so that
allocations route to this pool.
"""
def __init__(self, pool: MemPool):
super().__init__(pool)
@staticmethod
def active_pool() -> Optional[torch_npu._C._MemPool]:
r"""Returns the active MemPool"""
return torch_npu._C._MemPoolContext.active_pool()
@contextlib.contextmanager
def use_mem_pool(pool: MemPool, device=None):
r"""A context manager that routes allocations to a given pool.
Args:
pool(torch_npu.npu.MemPool): a MemPool object to be made active so that
allocations route to this pool.
device (torch.device or int, optional): selected device. Uses MemPool on
the current device, given by :func:`~torch_npu.npu.current_device,
if :attr:`device` is ``None`` (default).
"""
ctx = MemPoolContext(pool)
device_index = (
torch_npu.npu.current_device() if device is None else _get_device_index(device)
)
torch_npu._C._npu_beginAllocateToPool(device_index, pool.id)
try:
yield
finally:
torch_npu._C._npu_endAllocateCurrentStreamToPool(device_index, pool.id)
del ctx
def _record_memory_history(enabled="all", *args, **kwargs):
"""Enable recording of stack traces associated with memory
allocations, so you can tell what allocated any piece of memory in
:func:`torch.npu.memory._snapshot()`.
In addition too keeping stack traces with each current allocation and free,
this will also enable recording of a history of all alloc/free events.
Use :func:`torch.npu.memory._snapshot()` to retrieve this information,
and the tools in `_memory_viz.py` to visualize snapshots.
The Python trace collection is fast (2us per trace), so you may consider
enabling this on production jobs if you anticipate ever having to debug
memory issues.
C++ trace collection is also fast (~50ns/frame), which for many typical programs
works out to ~2us per trace, but can vary depending on stack depth.
Args:
enabled (Literal[None, "state", "all"], optional):
`None`, disable recording memory history.
`"state"`, keep information for currenly allocated memory.
`"all"`, additionally keep a history of all alloc/free calls.
Defaults to "all".
context (Literal[None, "state", "alloc", "all"], optional):
`None`, Do not record any tracebacks.
`"state"`, Record tracebacks for currently allocated memory.
`"alloc"`, additionally keep tracebacks for alloc calls.
`"all"`, additionally keep tracebacks for free calls.
Defaults to "all".
stacks (Literal["python", "all"], optional):
`"python"`, include Python, TorchScript, and inductor frames in tracebacks
`"all"`, additionally include C++ frames
Defaults to "all".
max_entries (int, optional): Keep a maximum of `max_entries`
alloc/free events in the recorded history recorded.
"""
return _record_memory_history_impl(enabled, *args, **kwargs)
def _record_memory_history_impl(
enabled: Optional[str] = "all",
context: Optional[str] = "all",
stacks: str = "all",
max_entries: int = sys.maxsize,
device=None,
):
torch_npu.npu._lazy_init()
torch_npu._C._npu_record_memory_history(enabled, context, stacks, max_entries)
def _snapshot(device=None):
"""Save a snapshot of NPU memory state at the time it was called.
The state is represented as a dictionary with the following structure.
.. code-block:: python
class Snapshot(TypedDict):
segments : List[Segment]
device_traces: List[List[TraceEntry]]
class Segment(TypedDict):
# Segments are memory returned from a aclrtMalloc call.
# The size of reserved memory is the sum of all Segments.
# Segments are cached and reused for future allocations.
# If the reuse is smaller than the segment, the segment
# is split into more then one Block.
# empty_cache() frees Segments that are entirely inactive.
address: int
total_size: int # aclrtMalloc'd size of segment
stream: int
segment_type: Literal['small', 'large'] # 'large' (>1MB)
allocated_size: int # size of memory in use
active_size: int # size of memory in use or in active_awaiting_free state
blocks : List[Block]
class Block(TypedDict):
# A piece of memory returned from the allocator, or
# current cached but inactive.
size: int
requested_size: int # size requested during malloc, may be smaller than
# size due to rounding
address: int
state: Literal['active_allocated', # used by a tensor
'active_awaiting_free', # waiting for another stream to finish using
# this, then it will become free
'inactive',] # free for reuse
frames: List[Frame] # stack trace from where the allocation occurred
class Frame(TypedDict):
filename: str
line: int
name: str
class TraceEntry(TypedDict):
# When `torch.npu.memory._record_memory_history()` is enabled,
# the snapshot will contain TraceEntry objects that record each
# action the allocator took.
action: Literal[
'alloc' # memory allocated
'free_requested', # the allocated received a call to free memory
'free_completed', # the memory that was requested to be freed is now
# able to be used in future allocation calls
'segment_alloc', # the caching allocator ask aclrtMalloc for more memory
# and added it as a segment in its cache
'segment_free', # the caching allocator called aclrtFree to return memory
# to npu possibly trying free up memory to
# allocate more segments or because empty_caches was called
'oom', # the allocator threw an OOM exception. 'size' is
# the requested number of bytes that did not succeed
'snapshot' # the allocator generated a memory snapshot
# useful to coorelate a previously taken
# snapshot with this trace
]
addr: int # not present for OOM
frames: List[Frame]
size: int
stream: int
device_free: int # only present for OOM, the amount of
# memory npu still reports to be free
Returns:
The Snapshot dictionary object
"""
return torch_npu._C._npu_memorySnapshot()
def _dump_snapshot(filename="dump_snapshot.pickle"):
"""
Save a pickled version of the `torch.memory._snapshot()` dictionary to a file.
This file can be opened by the interactive snapshot viewer at pytorch.org/memory_viz
Args:
filename (str, optional): Name of the file to create. Defaults to "dump_snapshot.pickle".
"""
s = _snapshot()
with os.fdopen(os.open(filename, os.O_WRONLY | os.O_CREAT, stat.S_IWUSR), "wb") as f:
pickle.dump(s, f)
device = torch_npu.npu.current_device()
save_success = torch_npu._C._npu_saveDevMemUsageInfo(device)
if not save_success:
prof_path = os.path.dirname(os.path.abspath(filename))
activities = {torch_npu.profiler.ProfilerActivity.CPU, torch_npu.profiler.ProfilerActivity.NPU}
torch_npu._C._profiler._init_profiler(prof_path, activities)
prof_config = [prof_path, False, True, False, False, False, torch_npu._C._profiler._ExperimentalConfig()]
npu_prof_config = torch_npu._C._profiler.NpuProfilerConfig(*tuple(prof_config))
torch_npu._C._profiler._start_profiler(npu_prof_config, activities)
torch_npu._C._profiler._stop_profiler()
torch_npu._C._profiler._finalize_profiler()
def _save_segment_usage(filename="output.svg", snapshot=None):
if snapshot is None:
snapshot = _snapshot()
with os.fdopen(os.open(filename, os.O_WRONLY | os.O_CREAT, stat.S_IWUSR), "w") as f:
f.write(_segments(snapshot))
def _save_memory_usage(filename="output.svg", snapshot=None):
if snapshot is None:
snapshot = _snapshot()
with os.fdopen(os.open(filename, os.O_WRONLY | os.O_CREAT, stat.S_IWUSR), "w") as f:
f.write(_memory(snapshot))
