Google Git
Sign in
android / platform / external / pytorch / 7ce932a92c / . / torch / fx / experimental / symbolic_shapes.py
blob: 00fb07be52e6ed8aa81ecea67fe061bc9dcfb8fd [file] [log] [blame]
import builtins
import collections
import functools
import inspect
import itertools
import logging
import math
import operator
import re
import sys
import textwrap
import threading
import traceback
from collections import defaultdict
from contextlib import contextmanager
from dataclasses import dataclass
from enum import Enum
from functools import lru_cache
from typing import cast, Callable, Dict, List, Optional, Set, Tuple, Type, Union
import torch
# NB: The sym_* functions are used via getattr() and must be imported here.
from torch import ( # noqa: F401
sym_float,
sym_max,
sym_min,
sym_not,
SymBool,
SymFloat,
SymInt,
)
from torch._guards import ShapeGuard, Source, TracingContext, detect_fake_mode
from torch.utils._sympy.interp import sympy_interp
from torch.utils._sympy.value_ranges import ValueRangeAnalysis, ValueRanges, ValueRangeError
from torch.utils._traceback import format_frame
from torch._utils_internal import signpost_event
InputList = List
DimList = List
SymTypes = (SymInt, SymFloat, SymBool)
log = logging.getLogger(__name__)
class GuardOnDataDependentSymNode(RuntimeError):
pass
import sympy
from sympy.printing.str import StrPrinter
from sympy.printing.precedence import precedence
from sympy.core.logic import fuzzy_and, fuzzy_or
aten = torch._ops.ops.aten # type: ignore[has-type]
__all__ = [
"has_symbolic_sizes_strides", "create_contiguous", "ShapeEnv", "is_concrete_int",
"SymDispatchMode", "FloorDiv", "guard_int", "guard_float", "guard_scalar", "wrap_node",
"method_to_operator", "hint_int", "SYMPY_INTERP", "free_symbols", "is_symbol_binding_fx_node",
"is_concrete_bool",
]
# These are modules that contain generic code for interacting with ShapeEnv
# which are unlikely to identify a particular interesting guard statement
@lru_cache(None)
def uninteresting_files():
import torch._inductor.sizevars
mods = [
sys.modules[__name__],
torch,
torch._inductor.sizevars,
]
return {inspect.getfile(m) for m in mods}
SYM_FUNCTION_MODE = None
# We don't bother with the metaclass as all of the dispatching logic happens
# entirely from Python
#
# Didn't bother with ancestors for now, unlikely to have multiple modes for
# symints right now
class ConstraintViolationError(RuntimeError):
pass
# SymDispatchMode gets invoked whenever an operation is processed on
# a PySymInt. When this occurs, you get called at __sym_dispatch__
# with the operation in question. This is symmetric to TorchDispatchMode
# but with some caveats:
#
# - In TorchDispatchMode, you get the same arguments as what a user
# invoked your API with; e.g., if you call torch.ops.aten.foo(a, b),
# you get (a, b) as args to your call. In SymDispatchMode, if
# you call a + b (where a and b are SymInts), you will get
# (a.node, b.node) as your args (these are PySymInts)
#
# - SymInt/PySymInt don't have FX proxy support (unlike, e.g., Tensor).
# So you have to manually call Tracer/create_node to write into
# the graph. See ProxySymDispatchMode for an example
#
class SymDispatchMode:
def __sym_dispatch__(self, func, types, args, kwargs):
raise NotImplementedError()
def __enter__(self):
global SYM_FUNCTION_MODE
old = SYM_FUNCTION_MODE
if hasattr(self, "inner"):
raise RuntimeError(f"{self} has already been used as a mode. Please use a fresh version")
else:
self.inner = old
SYM_FUNCTION_MODE = self
return self
def __exit__(self, exc_type, exc_val, exc_tb):
global SYM_FUNCTION_MODE
SYM_FUNCTION_MODE = self.inner
def has_symbolic_sizes_strides(elem):
return elem._has_symbolic_sizes_strides
def create_contiguous(shape):
strides = [1]
for dim in reversed(shape[:-1]):
strides.append(dim * strides[-1])
return list(reversed(strides))
def _handle_sym_dispatch(func, args, kwargs):
global SYM_FUNCTION_MODE
mode = SYM_FUNCTION_MODE
assert mode
SYM_FUNCTION_MODE = mode.inner
try:
# TODO: properly compute types
types: List[Type] = []
return mode.__sym_dispatch__(func, types, args, kwargs)
finally:
SYM_FUNCTION_MODE = mode
def hint_int(a):
if isinstance(a, torch.SymInt):
return a.node.require_hint()
assert type(a) is int, a
return a
def has_hint(a):
if isinstance(a, SymTypes):
return a.node.has_hint()
return True
def is_concrete_int(a: Union[int, SymInt]):
r""" Utility to check if underlying object
in SymInt is concrete value. Also returns
true if integer is passed in.
Args:
a (SymInt or int): Object to test if it int
"""
assert isinstance(a, (SymInt, int))
if isinstance(a, int):
return True
if isinstance(a.node.expr, sympy.core.numbers.Integer):
return True
return False
def is_concrete_bool(a: Union[bool, SymBool]):
r""" Utility to check if underlying object
in SymBool is concrete value. Also returns
true if integer is passed in.
Args:
a (SymBool or bool): Object to test if it bool
"""
assert isinstance(a, (SymBool, bool))
if isinstance(a, bool):
return True
if isinstance(a.node.expr, (sympy.logic.boolalg.BooleanTrue, sympy.logic.boolalg.BooleanFalse)):
return True
return False
# Returns True if every size dim on the tensor has a hint
# TODO: Should this include strides too? For now it doesn't matter,
# that's quite an obscure case
def tensor_has_hints(t):
return all(has_hint(s) for s in t.size())
def free_symbols(val: Union[SymInt, torch.Tensor]) -> Set[sympy.Symbol]:
if isinstance(val, (SymInt, SymFloat)):
return val.node.expr.free_symbols
elif isinstance(val, sympy.Expr):
return val.free_symbols
elif isinstance(val, (int, float, bool)):
return set()
elif isinstance(val, torch.Tensor):
return (
free_symbols(val.size()) |
free_symbols(val.stride()) |
free_symbols(val.storage_offset())
)
elif isinstance(val, (tuple, list)):
r = set()
for s in val:
r |= free_symbols(s)
return r
else:
raise AssertionError(f"cannot compute free_symbols of {val} {type(val)}")
# WARNING: Don't use this on Dynamo produced graphs, they don't have meta
# setup!
def is_symbol_binding_fx_node(node) -> Optional[sympy.Symbol]:
if (
node.op == "placeholder" and
"val" in node.meta and
isinstance(node.meta["val"], torch.SymInt) and
isinstance(node.meta["val"].node.expr, sympy.Symbol)
):
return node.meta["val"].node.expr
return None
def find_symbol_binding_fx_nodes(graph):
return {
node.meta["val"].node.expr: node
for node in graph.nodes
if is_symbol_binding_fx_node(node)
}
def definitely_true(a):
"""
Returns True only if we can tell that a is True, possibly introducing
a guard in the process. If a depends on some unbacked SymInt, we may
return False even though there may exist a possible value of the SymInt
that would cause the expression to return True.
When is it appropriate to use definitely_true? First, if you can use
a higher level combinator like parallel_or/parallel_and, prefer using
those instead, they are definitely safe (modulo short-circuiting).
Second, it can be used if the program would behave equivalently if
definitely_true always returned False (parallel_or/parallel_and are
examples of this pattern, modulo short-circuiting). Finally, it even
be OK if the program wouldn't behave equivalently, so long as the
change is semantics preserving. It can be semantics preserving if
the program errors in more cases than it did previously (but otherwise
behaves identically), or if it changes some quantity in a way that
doesn't matter (e.g., strides often fall in this bucket.)
"""
if isinstance(a, SymBool):
if a.node.has_hint():
return guard_bool(a)
else:
return False
return bool(a)
def definitely_false(a):
"""
Returns True only if we can tell that a is False, possibly introducing
a guard in the process. If a depends on some unbacked SymInt, we may
return False even though there may exist a possible value of the SymInt
that would cause the expression a to be False. See definitely_true
for more usage guidance.
"""
if isinstance(a, SymBool):
if a.node.has_hint():
return not guard_bool(a)
else:
return False
return not bool(a)
# TODO: could improve parallel_or/parallel_and by avoiding guards
# if there exists a quantity that can be handled un-guardedly. However,
# for backed SymInts, avoiding guards doesn't really matter in practice,
# so I chose not to do it.
def parallel_or(*args):
"""
Evaluate the logical OR of several arguments, avoiding guarding on
unbacked SymInts if another argument is definitely True.
"""
if any(definitely_true(args) for a in args):
return True
return any(args)
def parallel_and(*args):
"""
Evaluate the logical FALSE of several arguments, avoiding guarding on
unbacked SymInts if another argument is definitely False.
"""
if any(definitely_false(args) for a in args):
return False
return all(args)
def guard_scalar(a):
if isinstance(a, (SymBool, bool)):
return guard_bool(a)
elif isinstance(a, (SymInt, int)):
return guard_int(a)
elif isinstance(a, (SymFloat, float)):
return guard_float(a)
else:
raise AssertionError(f"unrecognized scalar {a}")
def _constrain_symbol_range(shape_env, s: sympy.Symbol, min: int, max: int):
if r := shape_env.var_to_range.get(s, None):
shape_env.var_to_range[s] = ValueRanges(
builtins.max(r.lower, min), builtins.min(r.upper, max)
)
else:
shape_env.var_to_range[s] = ValueRanges(min, max)
# inclusive both ways
def constrain_range(a, *, min: Optional[int], max: Optional[int] = None):
"""
Applies a constraint that the passed in SymInt must lie between min-max
inclusive-inclusive, WITHOUT introducing a guard on the SymInt (meaning
that it can be used on unbacked SymInts). If min/max are None, we assume
that the dimension is unbounded in that direction. Repeated application
of constrain_range intersects the ranges. This is a fairly low level API
that doesn't have a lot of safety guarantees (TODO: provide higher level
APIs).
Currently, we use this API in the following circumstance: when we allocate
an unbacked SymInt, denoting an integer quantity which is data dependent,
we ordinarily do not know anything about what values it may take. This
means that any sort of guard on it will immediately fail. However, in
many cases, we know something about the unbacked SymInt: for example, we
know that nonzero(x).size(0) must be >= 0. We use constrain_range to
narrow the possible range, declaring that negative symbols are impossible.
This permits to definitely answer True to queries like 'nnz >= 0', even if
we don't know what the actual (hinted) value of 'nnz' is. In fact, we
actually use constrain_range to unsoundly discharge common guards: for an
unbacked SymInt produced by nonzero, we will also assume that it is not
equal to 0/1 (even though these are perfectly possible values at runtime),
because we generally expect graphs that are valid for N=2 to also be valid
for N=1.
.. warning::
If you use constrain_range in the context of tracing, we do NOT check
that the constraint was actually valid at runtime! In fact, we
cannot (easily) do so, as we currently unsoundly assume that unbacked
SymInt can never be zero/one, even if it may actually take on these
values at runtime (we assume that a graph that is valid for N=2 will
also be valid for N=1).
"""
if min is None:
min = -sympy.oo
if max is None:
max = sympy.oo
if not isinstance(a, SymInt):
constrain_range_int(a, min=min, max=max)
return
if isinstance(a.node.expr, sympy.Integer):
if not (min <= int(a.node.expr) <= max):
raise ValueRangeError(f"Invalid value {int(a.node.expr)} for range [{min}:{max}]")
return
assert isinstance(a.node.expr, sympy.Symbol), "constraining non-Symbols NYI"
# TODO: Shouldn't we install a guard if the symbol is backed? Or is the
# semantics that this is an "unchecked" assert (but it this actually
# something useful? Might be better to restrict only for unbacked
# SymInt).
_constrain_symbol_range(a.node.shape_env, a.node.expr, min, max)
def constrain_range_int(a, *, min, max):
"""
Constrain range on concrete int value.
This can happens for the following scenarios:
- Eager mode execution and real int value is provided.
- During tracing the traced symbol is resolved as a static integer (see
PR #101655 for more details).
"""
assert not isinstance(a, SymInt)
if not (min <= a <= max):
raise ValueRangeError(f"Invalid value {a} for range [{min}:{max}]")
if (
(fake_mode := detect_fake_mode()) is not None and
getattr(fake_mode, "shape_env", None) is not None
):
# If we are tracing with a fake mode then add this integer to the
# shape_env's var_to_range
sym_integer = sympy.Integer(a)
shape_env = fake_mode.shape_env
_constrain_symbol_range(shape_env, sym_integer, min, max)
shape_env.var_to_stack[sym_integer] = TracingContext(fake_mode).extract_stack()
def constrain_unify(a, b):
"""
Given two SymInts, constrain them so that they must be equal. NB:
this will not work with SymInts that represent nontrivial expressions
(yet!)
"""
# TODO: Maybe dedupe this with _maybe_guard_eq?
if not isinstance(a, SymInt):
if not isinstance(b, SymInt):
assert a == b
else:
assert isinstance(b.node.expr, sympy.Symbol), "constraining non-Symbols NYI"
shape_env = b.node.shape_env
shape_env.replacements[b.node.expr] = sympy.Integer(a)
else:
# TODO: Actually, we can support this as long as one of them is a symbol.
# NB: We can't actually do "unification" as our operators are not
# injective
assert isinstance(a.node.expr, sympy.Symbol), "constraining non-Symbols NYI"
shape_env = a.node.shape_env
if not isinstance(b, SymInt):
shape_env.replacements[a.node.expr] = sympy.Integer(b)
else:
assert a.node.shape_env is b.node.shape_env
assert isinstance(b.node.expr, sympy.Symbol), "constraining non-Symbols NYI"
new_var = shape_env._find(a.node.expr)
shape_env.replacements[b.node.expr] = new_var
def guard_bool(a):
if isinstance(a, SymBool):
return a.node.guard_bool("", 0) # NB: uses Python backtrace
assert type(a) is bool, a
return a
def guard_int(a):
if isinstance(a, SymInt):
return a.node.guard_int("", 0) # NB: uses Python backtrace
assert type(a) is int, a
return a
def guard_float(a):
if isinstance(a, SymFloat):
return a.node.guard_float("", 0) # NB: uses Python backtrace
assert isinstance(a, float), a
return a
# Drop in replacement for math.sqrt
def sym_sqrt(a):
if hasattr(a, '__sym_sqrt__'):
return a.__sym_sqrt__()
return math.sqrt(a)
def to_node(self, num):
if isinstance(num, SymTypes):
return num.node
elif type(num) is bool:
return self.wrap_bool(num)
elif type(num) is int:
return self.wrap_int(num)
elif type(num) is float:
return self.wrap_float(num)
else:
# NotImplemented is important so that Python tries the
# other magic method
return NotImplemented
# Given a GraphModule, return all the FakeTensors for all the placeholders
def fx_placeholder_vals(gm):
return [n.meta['val'] for n in gm.graph.nodes if n.op == "placeholder"]
def fx_placeholder_targets(gm):
return [n.target for n in gm.graph.nodes if n.op == "placeholder"]
# Given a GraphModule and arguments to run it with, evaluate that the guards
# for its associated ShapeEnv are satisfied by the passed arguments. This
# WILL check for duck sizing.
def eval_guards(gm, *args, ignore_static=True):
return gm.shape_env.evaluate_guards_for_args(fx_placeholder_vals(gm), args, ignore_static=ignore_static)
def bind_symbols(gm, *args):
return gm.shape_env.bind_symbols(fx_placeholder_vals(gm), args)
class DimDynamic(Enum):
"""
Controls how to perform symbol allocation for a dimension. It is always
sound to default this to DYNAMIC, but the policies DUCK and STATIC can
result in better trace-time and compile-time performance, as they reduce
the number of allocated symbols and generally make your graph more static.
NB: If we notice you've applied a constraint to the dimension, we will
force it to DYNAMIC for simplicity.
DimDynamic is controlled by a variety of higher level UX features.
Currently:
- In eager mode, the default policy is DUCK.
- The default is changed to STATIC with assume_static_by_default.
- An individual dim is marked DYNAMIC if you mark_dynamic_dim.
- In export mode, the default policy is STATIC.
- An individual dim is marked DYNAMIC if you mention it as dynamic_dim
in the constraints kwarg.
"""
# Treat the dimension symbolically
DYNAMIC = 0
# Treat the dimension symbolically, but if its hint matches another
# dynamic dimension, unify the two symbols ("duck sizing")
DUCK = 1
# Treat the dimension statically based on its hint
STATIC = 2
# NB: These constraints affect both clients and backends: given some
# constraint C, the client must pass inputs that satisfy the constraint,
# while a backend must not introduce guards BEYOND this constraint.
# For clarity, we document the implications on both sides for both the client
# and the backend.
#
# NB: These constraints are on a *single* dimension. In principle, we could
# also have multi-dimension constraints, but our guess is that this is not
# actually useful and so we are not supporting it right now.
#
# NB: Strict constraints are typically only suitable for export, as in eager
# a backend like inductor may validly introduce extra, discretionary guards
# to improve performance of code. A StrictMinMaxConstraint would be brittle
# under future optimizations performed by inductor; we don't guarantee
# eager code with StrictMinMaxConstraint will keep working in the future!
@dataclass(frozen=True)
class Constraint:
warn_only: bool
@dataclass(frozen=True)
class StrictMinMaxConstraint(Constraint):
"""
For clients: the size at this dimension must be within 'vr' (which
specifies a lower and upper bound, inclusive-inclusive) AND it
must be non-negative and should not be 0 or 1 (but see NB below).
For backends: there must not be any guards on this dimension which
are not implied by the given lower and upper bound. Regardless of
the lower bound, the backend can assume the size is non-negative
and that it is not 0 or 1.
An unbounded StrictMinMaxConstraint can be thought of as a strict version
of "RelaxedUnspecConstraint".
NB: Export will often unsoundly assume that a graph works for 0/1, even
though at trace time we assumed size is not 0 or 1. The idea is that
if we produce a graph that works for a range of values, it will be OK
for N=0/1 too.
"""
vr: ValueRanges
def render(self, source: Source):
# TODO: better printing for -oo and oo
return f"{self.vr.lower} <= {source.name()} <= {self.vr.upper}"
@dataclass(frozen=True)
class RelaxedUnspecConstraint(Constraint):
"""
For clients: no explicit constraint; constraint is whatever is implicitly
inferred by guards from tracing.
For backends: there must exist at least TWO possible values for the
size at this dimension which satisfy the guards for this dimension.
In other words, this constraint helps us distinguish between "we don't
care if this dimension specializes or not" versus "this dimension must be
unspecialized." However, this constraint doesn't say very much about what
specialization is permitted; for example, if we guard on a size being
even, this would still be acceptable under an unspec constraint. This
makes RelaxedUnspecConstraint useful for eager mode, where your backend compiler
may add constraints to otherwise dynamic dimensions; we can't assert that
there are NO guards as this is brittle because compilers should be able to
add extra constraints. If you want to assert that there are no guards,
use StrictMinMaxConstraint with an unbounded ValueRanges.
"""
def render(self, source: Source):
return f"RelaxedUnspecConstraint({source.name()})"
# NB: None here indicates the client constraint is whatever is implicitly
# inferred by guards from tracing, and that a backend can add whatever guards
# it wants (including fully specializing the value).
DimConstraint = Union[StrictMinMaxConstraint, RelaxedUnspecConstraint, None]
@dataclass(frozen=True)
class EqualityConstraint(Constraint):
"""
Given pairs of sources corresponding to pairs of dynamic dimensions that
are specified equal, represent them in a union-find data structure so that
we can efficiently check whether two such sources are transitively equal.
"""
source_pairs: List[Tuple[Source, Source]]
def __post_init__(self):
object.__setattr__(self, "_parents", {})
for source1, source2 in self.source_pairs:
self._union(self._find(source1), self._find(source2))
def _find(self, source):
if source in self._parents:
return self._find(self._parents[source])
else:
return source
def _union(self, root1, root2):
if root1 != root2:
self._parents[root1] = root2
def render(self):
buf = ", ".join(
f"{source1.name()} == {source2.name()}"
for (source1, source2) in self.source_pairs
)
return "{" + buf + "}"
def is_equal(self, source1, source2):
return self._find(source1) == self._find(source2)
# TODO: An incomplete list
# 1. Set variables to be equal when we do equality
# 2. Specialize on 0/1 when we do subtraction
class SymNode:
"""
This is a type erased SymInt/SymFloat which we use to do actual operations.
End users don't touch this. Magic methods are NOT defined on this object.
"""
def __init__(self, expr, shape_env, pytype, hint: Optional[Union[int, float]], constant=None):
self._expr = expr
self.shape_env = shape_env
self.pytype = pytype
# What's the difference between hint and constant?
#
# - A constant is known to be invariant across invocations of the model;
# it will always be this value. We only really know this when we
# encounter an honest-to-goodness literal (when wrapping it into
# a SymNode, we set constant.) Most of the time, constant is None
#
# - A hint is a *particular* value from the particular run we are
# tracing, but it may vary the next time around. It's useful to
# keep this around, as if we need a concrete value from a SymNode,
# we will return the hint and guard on the expression that produced
# it giving the same hint next time around. The hint is not
# guaranteed to be set either: if you have an unbacked SymNode,
# there won't be any hint; it was the result of some tensor-dependent
# computation, but we don't know what it actually is because we
# haven't actually run the tensor computation.
#
# hint_expr is only set if we don't have a hint. When it is set, it
# contains the expression which contains the unbacked symnodes that,
# if constrained, would allow this expression to be hinted again.
if hint is None:
self._hint_expr = self.expr.xreplace(shape_env.var_to_val)
self._hint = None
self._update_hint() # check if the replacement actually was enough
else:
self._hint_expr = None
self._hint = hint
self.constant: Optional[Union[int, float, bool]] = constant
@property
def expr(self):
self._update_expr()
return self._expr
# Check if we have replacements hint_expr that would allow us to
# simplify it into a hint
def _update_hint(self):
if self._hint_expr.free_symbols <= self.shape_env.replacements.keys():
new_hint = self.shape_env.replace(self._hint_expr)
# NB: unification constraints could result in a replacement that
# doesn't actually solve the hint! Check for this.
if new_hint.free_symbols:
self._hint_expr = new_hint
return
self._hint = self.pytype(new_hint)
self._hint_expr = None
@property
def hint(self):
if self._hint is None:
self._update_hint()
return self._hint
def has_hint(self):
return self._hint is not None
def require_hint(self):
if self._hint is None:
self._update_hint()
if self._hint is None:
raise self.shape_env._make_data_dependent_error(self._hint_expr, self.expr)
else:
return self._hint
else:
return self._hint
def _update_expr(self):
self._expr = self.shape_env.replace(self._expr)
def is_int(self):
return self.pytype is int
def is_float(self):
return self.pytype is float
def is_bool(self):
return self.pytype is bool
def wrap_int(self, num):
assert type(num) is int
return SymNode(sympy.Integer(num), self.shape_env, int, num, constant=num)
def wrap_float(self, num):
assert type(num) is float
return SymNode(sympy.Float(num), self.shape_env, float, num, constant=num)
def wrap_bool(self, num):
assert type(num) is bool
return SymNode(sympy.true if num else sympy.false, self.shape_env, bool, num, constant=num)
def clone(self):
return self
def str(self):
return f"{self.expr}"
def __str__(self):
return self.str()
def __repr__(self):
return self.str()
# These methods call the metaprogrammed methods, they're hand written
# here so we get good stack traces
def add(self, other) -> "SymNode": # noqa: F811
return self._add(other) # type: ignore[attr-defined]
def sub(self, other) -> "SymNode": # noqa: F811
return self._sub(other) # type: ignore[attr-defined]
def mul(self, other) -> "SymNode": # noqa: F811
return self._mul(other) # type: ignore[attr-defined]
def mod(self, other) -> "SymNode": # noqa: F811
return self._mod(other) # type: ignore[attr-defined]
def pow(self, other) -> "SymNode": # noqa: F811
return self._pow(other) # type: ignore[attr-defined]
def and_(self, other) -> "SymNode": # noqa: F811
return self._and_(other) # type: ignore[attr-defined]
def or_(self, other) -> "SymNode": # noqa: F811
return self._or_(other) # type: ignore[attr-defined]
def truediv(self, other) -> "SymNode": # noqa: F811
return self._truediv(other) # type: ignore[attr-defined]
def floordiv(self, other) -> "SymNode": # noqa: F811
return self._floordiv(other) # type: ignore[attr-defined]
def sym_not(self) -> "SymNode": # noqa: F811
return self._sym_not() # type: ignore[attr-defined]
def eq(self, other) -> "SymNode": # noqa: F811
return self._eq(other) # type: ignore[attr-defined]
def ne(self, other) -> "SymNode": # noqa: F811
return self._ne(other) # type: ignore[attr-defined]
def gt(self, other) -> "SymNode": # noqa: F811
return self._gt(other) # type: ignore[attr-defined]
def lt(self, other) -> "SymNode": # noqa: F811
return self._lt(other) # type: ignore[attr-defined]
def le(self, other) -> "SymNode": # noqa: F811
return self._le(other) # type: ignore[attr-defined]
def ge(self, other) -> "SymNode": # noqa: F811
return self._ge(other) # type: ignore[attr-defined]
def floor(self) -> "SymNode": # noqa: F811
return self._floor() # type: ignore[attr-defined]
def sym_float(self) -> "SymNode": # noqa: F811
return self._sym_float() # type: ignore[attr-defined]
def sym_int(self) -> "SymNode": # noqa: F811
return self._sym_int() # type: ignore[attr-defined]
def ceil(self) -> "SymNode": # noqa: F811
return self._ceil() # type: ignore[attr-defined]
def neg(self) -> "SymNode": # noqa: F811
return self._neg() # type: ignore[attr-defined]
def sym_min(self, other) -> "SymNode": # noqa: F811
return self._sym_min(other) # type: ignore[attr-defined]
def sym_max(self, other) -> "SymNode": # noqa: F811
return self._sym_max(other) # type: ignore[attr-defined]
def sym_sqrt(self) -> "SymNode": # noqa: F811
return self._sym_sqrt() # type: ignore[attr-defined]
def is_contiguous(self, sizes, strides) -> "SymNode": # noqa: F811
return self._is_contiguous(sizes, strides) # type: ignore[attr-defined]
def is_channels_last_contiguous_2d(self, sizes, strides) -> "SymNode": # noqa: F811
return self._is_channels_last_contiguous_2d(sizes, strides) # type: ignore[attr-defined]
def is_channels_last_contiguous_3d(self, sizes, strides) -> "SymNode": # noqa: F811
return self._is_channels_last_contiguous_3d(sizes, strides) # type: ignore[attr-defined]
def is_channels_last_strides_2d(self, sizes, strides) -> "SymNode": # noqa: F811
return self._is_channels_last_strides_2d(sizes, strides) # type: ignore[attr-defined]
def is_channels_last_strides_3d(self, sizes, strides) -> "SymNode": # noqa: F811
return self._is_channels_last_strides_3d(sizes, strides) # type: ignore[attr-defined]
def is_non_overlapping_and_dense_indicator(self, sizes, strides) -> "SymNode": # noqa: F811
return self._is_non_overlapping_and_dense_indicator(sizes, strides) # type: ignore[attr-defined]
# Make C++ happy
def sym_or(self, other): # noqa: F811
return self.or_(other)
def sym_and(self, other): # noqa: F811
return self.and_(other)
def is_non_overlapping_and_dense(self, sizes, strides):
return self.is_non_overlapping_and_dense_indicator(sizes, strides).eq(to_node(self, 1)) # type: ignore[attr-defined]
def int_(self):
return self.guard_int("", 0) # NB: uses Python backtrace
# You can manually trigger a guard with this function
def guard_int(self, file, line):
# TODO: use the file/line for some useful diagnostic on why a
# guard occurred
r = self.shape_env.evaluate_expr(self.expr, self.hint)
try:
return int(r)
except Exception:
log.warning("Failed to convert to int: %s", r)
raise
def guard_float(self, file, line):
# TODO: use the file/line for some useful diagnostic on why a
# guard occurred
r = self.shape_env.evaluate_expr(self.expr, self.hint)
try:
return float(r)
except Exception:
log.warning("Failed to convert to float: %s", r)
raise
def guard_bool(self, file, line):
# TODO: use the file/line for some useful diagnostic on why a
# guard occurred
r = self.shape_env.evaluate_expr(self.expr, self.hint)
try:
return bool(r)
except Exception:
log.warning("Failed to convert to bool: %s", r)
raise
def bool_(self):
return self.guard_bool("", 0)
# Overloaded to be compatible with regular Python.
# https://github.com/pytorch/pytorch/issues/90900
class Pow(sympy.Function):
@classmethod
def eval(cls, base, exp):
if exp.is_zero:
return sympy.Integer(1)
elif base.is_zero and exp < 0:
raise ZeroDivisionError(f"{base} cannot be raised to a negative power")
else:
return base ** exp
# Overloaded to be compatible with regular Python.
# https://github.com/pytorch/pytorch/issues/90900
class TrueDiv(sympy.Function):
@classmethod
def eval(cls, base, divisor):
if divisor.is_zero:
raise ZeroDivisionError("division by zero")
else:
return base / divisor
class FloorDiv(sympy.Function):
"""
We maintain this so that:
1. We can use divisibility guards to simplify FloorDiv(a, b) to a / b.
2. Printing out the expression is nicer (compared to say, representing a//b as (a - a % b) / b)
"""
nargs = (2,)
precedence = 50 # precedence of mul # noqa: F811
# Default return type for SymPy assumptions.
# https://docs.sympy.org/latest/guides/assumptions.html#implementing-assumptions-handlers
is_real = True
@property
def base(self):
return self.args[0]
@property
def divisor(self):
return self.args[1]
def _sympystr(self, printer):
base = printer.parenthesize(self.base, self.precedence)
divisor = printer.parenthesize(self.divisor, self.precedence)
return f"({base}//{divisor})"
# SymPy assumptions based on argument types.
def _eval_is_real(self):
return fuzzy_or([self.base.is_real, self.divisor.is_real])
def _eval_is_integer(self):
return fuzzy_and([self.base.is_integer, self.divisor.is_integer])
# Automatic evaluation.
# https://docs.sympy.org/latest/guides/custom-functions.html#best-practices-for-eval
@classmethod
def eval(cls, base, divisor):
def check_supported_type(x):
if (x.is_integer is False and x.is_real is False and x.is_complex) or x.is_Boolean:
raise TypeError(
f"unsupported operand type(s) for //: "
f"'{type(base).__name__}' and '{type(divisor).__name__}'"
f", expected integer or real")
check_supported_type(base)
check_supported_type(divisor)
# We don't provide the same error message as in Python because SymPy
# makes it difficult to check the types.
if divisor.is_zero:
raise ZeroDivisionError("division by zero")
if base.is_zero:
return sympy.S.Zero
if base.is_integer and divisor == 1:
return base
if base.is_real and divisor == 1:
return sympy.floor(base)
if isinstance(base, sympy.Integer) and isinstance(divisor, sympy.Integer):
return base // divisor
if isinstance(base, (sympy.Integer, sympy.Float)) and isinstance(divisor, (sympy.Integer, sympy.Float)):
return sympy.floor(base / divisor)
if isinstance(base, FloorDiv):
return FloorDiv(base.args[0], base.args[1] * divisor)
if isinstance(base, sympy.Add):
for a in base.args:
gcd = sympy.gcd(a, divisor)
if gcd == divisor:
return FloorDiv(base - a, divisor) + a / gcd
gcd = sympy.gcd(base, divisor)
if gcd != 1:
return FloorDiv(
sympy.simplify(base / gcd), sympy.simplify(divisor / gcd)
)
# TODO: As an indicator, this != 0 implies == 1 (and vice versa).
# Because we do not have the ability to guard on the stride permutation
# at the moment, it is hard to make further inferences when this is true,
# as although we know the tensor is contiguous in *some* layout, we don't
# know which one (however, you could, for example, make the inference that
# reshaping this to a 1D tensor can be guard-free.)
class IsNonOverlappingAndDenseIndicator(sympy.Function):
is_integer = True
@classmethod
def eval(cls, *args):
assert len(args) % 2 == 0
dim = len(args) // 2
# TODO: it is possible to make progress evaluating this guard
# even if not all of the inputs are known. For example, a 2D
# tensor with non-0/1 sizes but strides (0, 1) is definitely
# false, because we know its numel > 1 but it's broadcasted
# in dim 0.
if all(isinstance(a, sympy.Integer) for a in args):
size_args = args[0:dim]
stride_args = args[dim:]
return eval_is_non_overlapping_and_dense(
[int(a) for a in size_args],
[int(a) for a in stride_args]
)
return None
IndicatorTypes = (IsNonOverlappingAndDenseIndicator,)
@lru_cache(256)
def safe_expand(r):
if hasattr(r, 'expand'):
try:
return sympy.expand(r)
except RecursionError:
log.warning("RecursionError in sympy.expand(%s)", r)
return r
else:
return r
# Methods that have a `__foo__` as well as `__rfoo__`
reflectable_magic_methods = {
'add': lambda a, b: a + b,
'sub': lambda a, b: a - b,
'mul': lambda a, b: a * b,
'mod': lambda a, b: a % b,
'pow': lambda a, b: Pow(a, b),
'and': lambda a, b: sympy.And(a, b),
'or': lambda a, b: sympy.Or(a, b),
'truediv': lambda a, b: TrueDiv(a, b),
'floordiv': lambda a, b: FloorDiv(a, b),
}
def error():
raise AssertionError("shouldn't be hit")
def get_debugging_stack(num_frames_to_cut=2):
# cut this frame and the caller's frame by default
return ''.join(traceback.format_list(traceback.extract_stack()[:-num_frames_to_cut]))
def floor_ceil_helper(a, fn):
if isinstance(a, sympy.Mul):
aa = a.args
if len(aa) == 2 and isinstance(aa[0], sympy.Float) and aa[1].is_integer:
coef = sympy.Integer(aa[0])
if aa[0] == coef: # structural equality test
return coef * aa[1]
if isinstance(a, sympy.Float) and a == sympy.Integer(a) or isinstance(a, sympy.Integer):
return sympy.Integer(a)
return fn(a)
def floor_impl(a):
return floor_ceil_helper(a, sympy.floor)
def ceil_impl(a):
return floor_ceil_helper(a, sympy.ceiling)
magic_methods = {
**reflectable_magic_methods,
'sym_not': lambda a: ~a,
'eq': lambda a, b: sympy.Eq(a, b),
'ne': lambda a, b: sympy.Ne(a, b),
'gt': lambda a, b: sympy.Gt(a, b),
'lt': lambda a, b: sympy.Lt(a, b),
'le': lambda a, b: sympy.Le(a, b),
'ge': lambda a, b: sympy.Ge(a, b),
'floor': floor_impl,
'sym_float': lambda a: a, # Cannot use sympy.Float(a) here, coz it expects python literals
'ceil': ceil_impl,
'neg': lambda a: -a,
'sym_min': lambda a, b: sympy.Min(a, b),
'sym_max': lambda a, b: sympy.Max(a, b),
'sym_sqrt': lambda a: sympy.sqrt(a),
}
sizes_strides_methods = {
# TODO: These could also be done with indicators, maybe it is better
# for reasoning to do it that way
'is_contiguous': lambda sizes, strides: sympy_is_contiguous(sizes, strides),
'is_channels_last_contiguous_2d': lambda sizes, strides: sympy_is_channels_last_contiguous_2d(sizes, strides),
'is_channels_last_contiguous_3d': lambda sizes, strides: sympy_is_channels_last_contiguous_3d(sizes, strides),
'is_channels_last_strides_2d': lambda sizes, strides: sympy_is_channels_last_strides_2d(sizes, strides),
'is_channels_last_strides_3d': lambda sizes, strides: sympy_is_channels_last_strides_3d(sizes, strides),
'is_non_overlapping_and_dense_indicator': lambda sizes, strides: IsNonOverlappingAndDenseIndicator(*sizes, *strides),
}
alternate_impl_if_hinted_methods = {
"sym_min": builtins.min,
"sym_max": builtins.max,
}
def sympy_is_contiguous_generic(sizes, strides, dim_order):
dim = len(sizes)
if len(dim_order) != dim:
return sympy.false
is_contiguous = sympy.true
z = sympy.Integer(1)
# Contiguous if the strides make sense (or the dim is size 1)
for d in dim_order:
is_contiguous &= sympy.Eq(sizes[d], sympy.Integer(1)) | sympy.Eq(strides[d], z)
z *= sizes[d]
# OR if any size is zero
for d in range(dim):
is_contiguous |= sympy.Eq(sizes[d], sympy.Integer(0))
return is_contiguous
def sympy_is_contiguous(sizes, strides):
dim = len(sizes)
return sympy_is_contiguous_generic(sizes, strides, list(range(dim - 1, -1, -1)))
# NB: There is a TODO in C++ to allow omitting the batch dim. If that
# happens you will need to refactor this
def sympy_is_channels_last_contiguous_2d(sizes, strides):
return sympy_is_contiguous_generic(sizes, strides, [1, 3, 2, 0])
def sympy_is_channels_last_contiguous_3d(sizes, strides):
return sympy_is_contiguous_generic(sizes, strides, [1, 4, 3, 2, 0])
def sympy_is_channels_last_strides_generic(sizes, strides, dim_order):
dim = len(sizes)
if dim != len(dim_order):
return sympy.false
m = sympy.Integer(0)
r = sympy.true
# special case for trivial C dimension. default to NCHW
r &= sympy.Ne(strides[1], 0)
for d in dim_order:
r &= sympy.Ne(sizes[d], 0) & (strides[d] >= m)
# Fallback to NCHW as default layout for ambiguous cases
# This is the flaw of implicit memory_format from strides.
# N111 tensor with identical strides for size 1 dimension;
# Two cases could lead us here:
# a. N111 contiguous Tensor ([N,1,1,1]@[1,1,1,1])
# b. N11W contiguous Tensor sliced on the W-dimension.
# ([N,1,1,1]@[W,W,W,W])
if d == 0:
r &= sympy.Ne(m, strides[1])
# This is necessary to:
# 1. distinguish the memory_format of N1H1;
# [H, 1, 1, 1] channels_last stride
# [H, H, 1, 1] contiguous stride
# 2. permutation of 1C1W:
# [1, C, 1, H]@[HC, H, H, 1] transpose(1, 3)
# [1, H, 1, C]@[HC, 1, H, H] shouldn't be identified as
# channels_last
m = strides[d] * sympy.Max(sizes[d], 1)
return r
def sympy_is_channels_last_strides_2d(sizes, strides):
return sympy_is_channels_last_strides_generic(sizes, strides, [1, 3, 2, 0])
def sympy_is_channels_last_strides_3d(sizes, strides):
return sympy_is_channels_last_strides_generic(sizes, strides, [1, 4, 3, 2, 0])
# TODO: Deduplicate this with torch/_prims_common/__init__.py
def eval_is_non_overlapping_and_dense(sizes, strides):
return int(guard_bool(_eval_is_non_overlapping_and_dense(sizes, strides)))
def _eval_is_non_overlapping_and_dense(sizes, strides):
dim = len(sizes)
# Short-circuits for tensors of rank one, which are
# non-overlapping and "dense" if their stride is one
# or it is a 0/1 element tensor
if dim == 1:
return strides[0] == 1 or sizes[0] < 2
# Checks that there exists a permutation of the strides s.t. the tensor would be contiguous
# Sorts (length, stride) pairs by stride
lengths_and_strides = sorted(
zip(sizes, strides), key=operator.itemgetter(1)
)
# Unlike the C++ code, we don't move the 0/1 size dimensions to the
# end. So we have to keep going for this code.
expected_stride = 1
for length, stride in lengths_and_strides:
if length == 1:
continue
if stride != expected_stride:
return False
expected_stride *= length
return True
unary_magic_methods = {
'sym_float',
'ceil',
'floor',
'neg',
'sym_sqrt',
'sym_not',
}
bool_magic_methods = {"and", "or", "sym_not"}
magic_methods_on_math = {"ceil", "floor"}
magic_methods_on_submodule = {"sym_float", "sym_sqrt", "sym_min", "sym_max", "sym_not"}
magic_methods_on_operator_with_trailing_underscore = {"and", "or"}
def method_to_operator(method):
if method in magic_methods_on_operator_with_trailing_underscore:
method_attr = f"{method}_"
else:
method_attr = method
if method in magic_methods_on_submodule:
op = getattr(torch.fx.experimental.symbolic_shapes, method_attr)
elif method in magic_methods_on_math:
op = getattr(math, method_attr)
else:
op = getattr(operator, method_attr)
return op
SYMPY_INTERP = {
'Eq': operator.eq,
'Ne': operator.ne,
'Gt': operator.gt,
'Lt': operator.lt,
'Le': operator.le,
'Ge': operator.ge,
'Min': min,
'Max': max,
'Mod': operator.mod,
'FloorDiv': operator.floordiv,
'TrueDiv': operator.truediv,
'IsNonOverlappingAndDenseIndicator': eval_is_non_overlapping_and_dense,
'floor': math.floor,
'ceiling': math.ceil,
}
always_float_magic_methods = {"truediv", "sym_float", "sym_sqrt", "pow"}
always_int_magic_methods = {"ceil", "floor"}
always_bool_magic_methods = {"eq", "ne", "gt", "lt", "le", "ge", "and", "or", "sym_not", "is_non_overlapping_and_dense"}
def wrap_node(x):
# TODO: let C++ also take advantage of this
if isinstance(x, SymNode) and x.constant is not None:
return x.constant
if x.is_int():
return SymInt(x)
elif x.is_float():
return SymFloat(x)
elif x.is_bool():
return SymBool(x)
else:
raise AssertionError(f"unrecognized return type {x}")
def _make_node_magic(method, func):
func = lru_cache(256)(func)
if method in magic_methods_on_operator_with_trailing_underscore:
method_attr = f"{method}_"
else:
method_attr = method
def binary_magic_impl(self, other):
op = method_to_operator(method)
out_hint = None
if self.hint is not None and other.hint is not None:
out_hint = op(self.hint, other.hint)
alternate_impl = alternate_impl_if_hinted_methods.get(method)
if alternate_impl and out_hint is not None:
return to_node(self, alternate_impl(wrap_node(self), wrap_node(other)))
if SYM_FUNCTION_MODE:
return to_node(self, _handle_sym_dispatch(op, (wrap_node(self), wrap_node(other)), {}))
assert isinstance(other, SymNode)
other_expr = other.expr
# TODO: consider constant prop here
expr = self.shape_env.replace(self.expr)
other_expr = self.shape_env.replace(other_expr)
try:
out = func(expr, other_expr)
except Exception:
log.warning("failed to eval %s(%s, %s)", method, expr, other_expr)
raise
out = safe_expand(out)
pytype: Type
# This is not strictly correct. In Python, a**b may return complex when
# a < 0 and b is a float: (-1)**2.1. Same for sympy.sqrt(-3.14). This
# returns a float while both arguments are ints: 2**(-1). Also, max and
# min do not type promote. To avoid having data-dependent control flow
# here, we just set the type to float if one of the args is a float. In
# case of a type mismatch, we assume that it will be detected during
# evaluation.
if method in always_float_magic_methods:
pytype = float
elif method in always_bool_magic_methods:
pytype = bool
elif self.pytype is float or other.pytype is float:
pytype = float
else:
pytype = self.pytype
return SymNode(out, self.shape_env, pytype, out_hint)
def unary_magic_impl(self):
op = method_to_operator(method)
if SYM_FUNCTION_MODE:
return to_node(self, _handle_sym_dispatch(op, (wrap_node(self),), {}))
# TODO: consider constant prop here
expr = self.shape_env.replace(self.expr)
if method == "floor" or method == "ceiling":
expr = self.shape_env._simplify_floor_div(expr)
try:
out = func(expr)
except Exception:
log.warning("failed to eval %s(%s)", method, expr)
raise
out_hint = None
if self.hint is not None:
out_hint = op(self.hint)
out = safe_expand(out)
pytype: Type
if method in always_int_magic_methods:
pytype = int
elif method in always_float_magic_methods:
pytype = float
else:
pytype = self.pytype
return SymNode(out, self.shape_env, pytype, out_hint)
if method in unary_magic_methods:
setattr(SymNode, f"_{method_attr}", unary_magic_impl)
else:
setattr(SymNode, f"_{method_attr}", binary_magic_impl)
def _make_node_sizes_strides(method, func):
# NB: don't LRU cache, lots of arguments
def sizes_strides_impl(self, sizes, strides):
op = getattr(sys.modules[__name__], method)
if SYM_FUNCTION_MODE:
return to_node(
self,
_handle_sym_dispatch(
op,
([wrap_node(s) for s in sizes], [wrap_node(s) for s in strides]),
{}
)
)
size_exprs = [s.expr for s in sizes]
stride_exprs = [s.expr for s in strides]
try:
out = func(size_exprs, stride_exprs)
except Exception:
log.warning("failed to eval %s(%s, %s)", method, size_exprs, stride_exprs)
raise
# bool is never expandable
size_hints = []
out_hint = None
for s in sizes:
if s.hint is None:
break
size_hints.append(s.hint)
else:
stride_hints = []
for s in strides:
if s.hint is None:
break
stride_hints.append(s.hint)
else:
out_hint = op(size_hints, stride_hints)
# NB: This is the indicator function, not the actual bool!
pytype: Type
if method.endswith("_indicator"):
pytype = int
else:
pytype = bool
return SymNode(out, self.shape_env, pytype, out_hint)
setattr(SymNode, f"_{method}", sizes_strides_impl)
# TODO: This is technically hotpath, but in the ideal end state
# guards on this will resolve at a higher level so you never
# spend time in this code
def sizes_strides_user(sizes, strides):
for a in itertools.chain(sizes, strides):
if isinstance(a, SymInt):
return wrap_node(getattr(a.node, method)(
[to_node(a.node, b) for b in sizes],
[to_node(a.node, b) for b in strides],
))
if method == "is_non_overlapping_and_dense_indicator":
return eval_is_non_overlapping_and_dense(sizes, strides)
else:
# TODO: this is an awful implementation
return bool(func(
[sympy.sympify(a) for a in sizes],
[sympy.sympify(a) for a in strides],
))
# Skip for is_non_overlapping_and_dense_indicator
if not hasattr(sys.modules[__name__], method):
setattr(sys.modules[__name__], method, sizes_strides_user)
for method, func in magic_methods.items():
_make_node_magic(method, func)
for method, func in sizes_strides_methods.items():
_make_node_sizes_strides(method, func)
def _make_user_magic(method, user_type):
# User magic takes care of wrapping the other operand into a node,
# so that our internal logic can assume everything is nodes
if method in magic_methods_on_operator_with_trailing_underscore:
method_attr = f"{method}_"
else:
method_attr = method
def unary_magic_impl(self):
return wrap_node(getattr(self.node, method_attr)())
def binary_magic_impl(self, other):
other_node = to_node(self.node, other)
if other_node is NotImplemented:
return NotImplemented
return wrap_node(getattr(self.node, method_attr)(other_node))
def rbinary_magic_impl(self, other):
other_node = to_node(self.node, other)
if other_node is NotImplemented:
return NotImplemented
return wrap_node(getattr(other_node, method_attr)(self.node))
if method in unary_magic_methods:
setattr(user_type, f"__{method}__", unary_magic_impl)
else:
setattr(user_type, f"__{method}__", binary_magic_impl)
if method in reflectable_magic_methods:
setattr(user_type, f"__r{method}__", rbinary_magic_impl)
for method, func in magic_methods.items():
if method in bool_magic_methods:
_make_user_magic(method, SymBool)
else:
_make_user_magic(method, SymInt)
_make_user_magic(method, SymFloat)
del method
del func
def _lru_cache(fn, maxsize=None):
"""
Wrapper around lru_cache that clears when new info about shapes has been
updated.
Use lru_cache if the output is always the same, regardless of the
constraints we know now (i.e. evaluate_expr)
Use _lru_cache otherwise.
"""
fn_cache = lru_cache(maxsize)(fn)
prior_key = None
@functools.wraps(fn)
def wrapper(self, *args, **kwargs):
nonlocal prior_key
if prior_key != self._get_key():
prior_key = self._get_key()
fn_cache.cache_clear()
return fn_cache(self, *args, **kwargs)
wrapper.cache_info = fn_cache.cache_info # type: ignore[attr-defined]
return wrapper
class ShapeGuardPrinter(StrPrinter):
def __init__(
self,
symbol_to_source,
source_ref,
var_to_sources,
):
super().__init__()
self.symbol_to_source = symbol_to_source
self.source_ref = source_ref
self.var_to_sources = var_to_sources
def _print_Symbol(self, expr) -> str:
assert isinstance(expr, sympy.Symbol), str(type(expr))
def repr_symbol_to_source():
return repr({
symbol: [s.name() for s in sources]
for symbol, sources in self.symbol_to_source.items()
})
assert self.symbol_to_source.get(expr), (
f"{expr} (could be from {[s.name() for s in self.var_to_sources[expr]]}) "
f"not in {repr_symbol_to_source()}. If this assert is failing, it could be "
"due to the issue described in https://github.com/pytorch/pytorch/pull/90665"
)
return self.source_ref(self.symbol_to_source[expr][0])
class LoggingShapeGuardPrinter(ShapeGuardPrinter):
def __init__(self, var_to_sources):
super().__init__(var_to_sources, lambda n: n.name(), var_to_sources)
class DynamicDimConstraintPrinter(StrPrinter):
"""
Printer for dynamic dim constraints.
- Instead of t.size()[d] it prints dynamic_dim(t, d)
- Instead of Eq(_, _), Mod(_, _), etc. it prints _ == _, _ % _, etc.
We use this to suggest code for specifying dynamic dim constraints.
"""
def __init__(self, symbol_to_source):
super().__init__()
self.symbol_to_source = symbol_to_source
def print_source(self, source) -> str:
return f"dynamic_dim({source.base.name()}, {source.idx})"
def _print_Symbol(self, expr) -> str:
assert isinstance(expr, sympy.Symbol), str(type(expr))
return self.print_source(self.symbol_to_source[expr][0])
def _print_Relational(self, expr):
return '%s %s %s' % (
self.parenthesize(expr.lhs, precedence(expr)),
expr.rel_op,
self.parenthesize(expr.rhs, precedence(expr))
)
class DimConstraints:
"""
Custom solver for a system of constraints on symbolic dimensions.
Solutions are "static" values or simplified "dynamic" constraints.
"""
def __init__(self, symbol_to_source, var_to_val, marked_dynamic):
# We try to solve systems of inequalities with 1 free variable.
self._univariate_inequalities: Dict[sympy.Symbol, Set[sympy.Expr]] = defaultdict(set)
# Among them, we prioritize solving for a free variable that has equalities.
# NOTE: _symbols_with_equalities is always a subset of _univariate_inequalities.keys()
# and removing a symbol from the former => removing it from the latter.
self._symbols_with_equalities: Set[sympy.Symbol] = set()
# A solution of a free variable with equalities becomes a substitution.
# We use these substitutions to simplify other constraints.
# NOTE: removing a symbol from _symbols_with_equalities => adding it to _substitutions.
self._substitutions: Dict[sympy.Symbol, sympy.Integer] = {}
# In general, constraints may have // and % operations.
# Of course, // can be expressed in terms of / and %.
# Our inequality solver can handle / but not %. So we need to transform them away.
# We do so by using the values of variables as hints to evaluate %.
# For soundness we record additional congruence guards and solve them separately.
self._var_to_val: Dict[sympy.Symbol, sympy.Integer] = var_to_val
self._congruences: Set[sympy.Expr] = defaultdict(set)
# We do not try to (directly) solve inequalities with > 1 free variables.
# NOTE: free variables in these inequalities cannot also be in _substitutions.
self._multivariate_inequalities: Set[sympy.Expr] = set()
# We park external equalities between free variables here.
self._symbolic_equivalences: List[Tuple[Source, sympy.Expr]] = []
# Solutions come in two forms:
# - (static) specializations
# - (dynamic) inequalities / congruences
self._static_results: Set[str] = set()
self._dynamic_results: Set[str] = set()
# printer for solutions
self._dcp = DynamicDimConstraintPrinter(symbol_to_source)
# inconsistencies found on substituting with concrete values / static solutions
self._inconsistencies: List[str] = []
# symbols that are marked dynamic
self._marked_dynamic = marked_dynamic
def rewrite_with_congruences(self, s, expr):
"""
Eliminate expressions of the form b // d and b % d while adding congruences of the form b % d == k.
This leaves rational operators (in particular of the form b / d) that our inequality solver can handle.
We solve the added congruences separately (using our congruence solver, see below).
"""
def mod_handler(*args):
# Suppose that we have an expression of the form b % d with free variable s.
# Using the value of s as a "hint," we can evaluate b % d to a value k.
# Then we can rewrite b % d to k while adding the guard b % d == k.
# NOTE(avik): This abstraction is provably sound but, in general, incomplete. It is complete IFF
# the original expression always evaluates to a constant value (i.e., it does not vary with s).
# In other words,
# - solutions of s with the rewritten expression are guaranteed to also be solutions of s with
# the original expression;
# - while it may be possible to find solutions of s with the original expression that are not
# solutions with the rewritten expression, in that case the original expression cannot evaluate
# to the same value for all solutions of s.
#
# Should we be worried about this incompleteness? No, because of the following reasons:
# 1. It unblocks dramatic simplification that would not be otherwise possible with current tech
# (i.e., "don't let perfect be the enemy of the good").
# 2. We already have a tradition of using hints to add guards in the compiler for making progress.
# 3. We have not yet seen a counterexample arise in practice! In particular, any congruence guards
# we generate (or simplify to) seem to be of the form b % d == k where k is a constant.
#
# Here's a theoretical counterexample: 3*s % (s + 1) == s - 2, that is satisfied by all s >= 2.
# With any hint (say) s = k, we'd rewrite this to: 3*s % (s + 1) == k - 2. But, substituting, we
# would then get k - 2 == s - 2, and thus s = k as the (only, constant) solution!
base, divisor = args
base, divisor = self.rewrite_with_congruences(s, base), self.rewrite_with_congruences(s, divisor)
mod_reduced = base.subs(self._var_to_val) % divisor.subs(self._var_to_val)
congruence = (base - mod_reduced) % divisor
if congruence != 0:
self._congruences[s].add(congruence)
return mod_reduced
def floor_div_handler(*args):
# Suppose that we have an expression of the form b // d with free variable s.
# Using the value of s, we can evaluate b % d to a value k.
# Then we can rewrite b // d to (b - k) / d, while adding the guard b % d == k.
# NOTE(avik): This is exactly equivalent to rewriting b // d as (b - (b % d)) / d
# and eliminating b % d as above.
base, divisor = args
base, divisor = self.rewrite_with_congruences(s, base), self.rewrite_with_congruences(s, divisor)
mod_reduced = base.subs(self._var_to_val) % divisor.subs(self._var_to_val)
congruence = (base - mod_reduced) % divisor
if congruence != 0:
self._congruences[s].add(congruence)
return (base - mod_reduced) / divisor
if expr.has(sympy.Mod):
expr = expr.replace(sympy.Mod, mod_handler)
if expr.has(FloorDiv):
expr = expr.replace(FloorDiv, floor_div_handler)
return expr
def add(self, expr):
if expr == sympy.true:
return
orig_expr = expr
orig_reduced = orig_expr.subs(self._var_to_val)
# TODO(avik): https://github.com/pytorch/pytorch/issues/101093
# It is possible that `expr` will fail the consistency check because of
# precision errors. Specifically, on substituting its free symbols with
# their concrete values, we might end up comparing floats. Until we have
# a fix for this issue, we delay raising such failures. See solve().
if orig_reduced == sympy.false:
self._inconsistencies.append(f"{orig_expr} is inconsistent!")
free_symbols = expr.free_symbols
assert free_symbols, f"Did not expect constraint with no free variables: {expr}"
if len(free_symbols) > 1:
# multivariate: record and move on
self._multivariate_inequalities.add(expr)
else:
# univariate: can solve these immediately
s = next(iter(free_symbols))
# eliminate // and % (see documentation of `rewrite_with_congruences` above)
expr = self.rewrite_with_congruences(s, expr)
if expr != sympy.true:
reduced = expr.subs(self._var_to_val)
if reduced == sympy.false:
self._inconsistencies.append(
f"{expr}, obtained by rewriting {orig_expr} with congruences, "
"is inconsistent!"
)
if isinstance(expr, sympy.Eq):
# special status for symbols that have equalities (see `solve` below)
self._symbols_with_equalities.add(s)
self._univariate_inequalities[s].add(expr)
def add_equality(self, source, expr):
if expr.free_symbols:
# these will resolve to either specializations or dynamic equality constraints
self._symbolic_equivalences.append((source, expr))
else:
# specialization, right here
self._static_results.add(f"{source.name()} == {expr}")
def reduce_congruences(self):
reduced_congruences = {}
for s, congruences in self._congruences.items():
remainder_modulus_pairs = []
congruences_to_check = set()
for congruence in congruences:
base, divisor = congruence.args
# We are given a congruence of the form base % divisor == 0 with a free variable s. So:
# - we transform this into an equation of the form base = divisor * tmp;
# - we solve this equation for s to get a linear solution with free variable tmp.
tmp = sympy.Symbol("tmp", integer=True)
symbol, solution = sympy.solve_linear(base - divisor * tmp, symbols=[s])
# See https://docs.sympy.org/latest/modules/solvers/solvers.html#sympy.solvers.solvers.solve_linear
# for how to interpret the results.
if s == symbol:
# This means the solution is of the form s = modulus*tmp + remainder.
modulus, remainder = sympy.polys.polytools.div(solution, tmp)
if isinstance(modulus, sympy.Integer) and isinstance(remainder, sympy.Integer):
# Make sure 0 <= remainder <= modulus.
remainder = remainder % modulus
remainder_modulus_pairs.append((remainder, modulus))
continue
# This means that we did not get a unique solution to the equation.
# No problem, we will check it.
congruences_to_check.add(congruence)
# Finally we solve for a congruence s such that s = r_i mod m_i for each (r_i, m_i).
# The solution will be a congruence of the form s = r mod m.
# NOTE(avik): Since the given m_i may not be pairwise coprime, we can't just use CRT.
if remainder_modulus_pairs:
remainder, modulus = sympy.ntheory.modular.solve_congruence(*remainder_modulus_pairs)
reduced_congruences[s] = {(s - remainder) % modulus}
substitution = {s: modulus * sympy.Symbol("tmp", integer=True) + remainder}
reduced_congruences[s].update(
congruence for congruence in congruences_to_check
if not sympy.checksol(congruence, substitution)
)
else:
reduced_congruences[s] = congruences_to_check
return reduced_congruences
def raise_inconsistencies(self):
if self._inconsistencies:
msg = "\n".join(self._inconsistencies)
self._inconsistencies.clear()
raise ValueError(f"The following inconsistencies were found:\n{msg}")
def _force_specialization(self, s):
val = self._var_to_val[s]
self._static_results.add(f"{self._dcp.symbol_to_source[s][0].name()} == {val}")
self._substitutions[s] = val
def specialize_divisor_symbols(self):
for expr in self._multivariate_inequalities:
for atom in expr.atoms(FloorDiv, sympy.Mod):
_, divisor = atom.args
for s in divisor.free_symbols:
self._force_specialization(s)
multivariate_inequalities = self._multivariate_inequalities
self._multivariate_inequalities = set()
for expr in multivariate_inequalities:
self.add(expr.subs(self._substitutions))
self.raise_inconsistencies()
self._univariate_inequalities = {
s: exprs
for s, exprs in self._univariate_inequalities.items()
if s not in self._substitutions
}
self._congruences = {
s: congruences
for s, congruences in self._congruences.items()
if s not in self._substitutions
}
def solve(self, disable_congruences=True):
self.raise_inconsistencies()
# as long as there are symbols with equalities, solve for them
# NOTE(avik): this is guaranteed to terminate (#iterations <= #symbols)
while(self._symbols_with_equalities):
s = self._symbols_with_equalities.pop()
exprs = self._univariate_inequalities.pop(s)
solution = sympy.solvers.inequalities.reduce_inequalities(exprs, s)
if isinstance(solution, sympy.And):
solution = next((arg for arg in solution.args if isinstance(arg, sympy.Eq)), solution)
assert isinstance(solution, sympy.Eq), f"Expected an equality constraint for {s}, got {solution}"
symbol, val = solution.args
assert symbol == s, f"Expected a constraint on {s} instead of on {symbol}"
# because this is univariate, the solution is a specialization
self._static_results.add(f"{self._dcp.symbol_to_source[s][0].name()} == {val}")
# add this as a substitution to simplify other constraints
self._substitutions[s] = val
# simplify multivariate inequalities: some of them will now become univariate!
multivariate_inequalities = self._multivariate_inequalities
self._multivariate_inequalities = set()
for expr in multivariate_inequalities:
self.add(expr.subs(s, self._substitutions[s]))
self.raise_inconsistencies()
self.specialize_divisor_symbols()
# solve linear congruences
# NOTE(avik): We do not need to solve them for symbols that have already been specialized.
reduced_congruences = self.reduce_congruences()
for s, congruences in reduced_congruences.items():
for congruence in congruences:
# any congruence that cannot be checked becomes a dynamic constraint as well
if s not in self._substitutions or not sympy.checksol(congruence, {s: self._substitutions[s]}):
if disable_congruences:
self._force_specialization(s)
self._univariate_inequalities.pop(s, None)
else:
self._dynamic_results.add(self._dcp.doprint(sympy.Eq(congruence, 0)))
# remaining symbols have only pure inequalities (no equalities)
for s, exprs in self._univariate_inequalities.items():
try:
solution = sympy.solvers.inequalities.reduce_inequalities(exprs, s)
# because this is univariate, the solution is a dynamic (range) constraint
if isinstance(solution, sympy.And):
for arg in solution.args:
self._dynamic_results.add(self._dcp.doprint(arg))
else:
self._dynamic_results.add(self._dcp.doprint(solution))
except NotImplementedError as e:
log.warning("Failed to reduce inequalities: %s", e)
for expr in exprs:
self._dynamic_results.add(self._dcp.doprint(expr))
# simplify symbolic equivalences: some of them will now become specializations!
symbolic_equivalences = self._symbolic_equivalences
self._symbolic_equivalences = []
for source, expr in symbolic_equivalences:
self.add_equality(source, expr.subs(self._substitutions))
# remaining symbolic equivalences become dynamic equality constraints
for source, expr in self._symbolic_equivalences:
self._dynamic_results.add(f"{self._dcp.print_source(source)} == {self._dcp.doprint(expr)}")
def forced_specializations(self):
return "\n".join([
(
f"\t{self._dcp.symbol_to_source[s][0].name()}, which was marked dynamic, "
f"must be specialized to {val}."
)
for s, val in self._substitutions.items()
if s in self._marked_dynamic
])
def prettify_results(self, original_signature: inspect.Signature):
# Note: Model inputs are wrapped as LocalSource in dynamo.
# LocalSource.name() wraps the name with L[""]. We use regular
# expression to do the replacement to avoid traversing up
# the source hierarchy manually.
def extract_and_rewrite_local(dc):
match = re.search(r"L\['(.+?)'\]", dc)
if match is None:
return
arg = match.expand(r'\1')
dc = re.sub(r"L\['(.+?)'\]", r'\1', dc)
return arg, dc
def group(results, args_index):
groups = defaultdict(list)
for dc in results:
local = extract_and_rewrite_local(dc)
if local is None:
# This can happen, e.g., with `assume_constant_result`.
# In that case, we drop the constraint.
# TODO(avik) Maybe we should generate an assertion here?
continue
arg, dc = local
if arg in args_index:
groups[args_index[arg]].append(dc)
else:
# This can happen, e.g., with decorators that change the signature.
# In that case, we drop the constraint. Seems hard to do better. :/
# TODO(avik) Maybe warn that `arg` in not in `signature`?
continue
sorted_groups = []
for idx, dcs in sorted(groups.items()):
_, arg = idx
sorted_groups.append((arg, sorted(dcs)))
return sorted_groups
# Instead of 2 <= dynamic_dim(...) simply suggest dynamic_dim(...).
# There is no change in behavior since 2 is the default lower bound.
def remove_default_lower_bound(dc):
return re.sub(r"2 <= dynamic_dim(.+)", r"dynamic_dim\1", dc)
signature = original_signature.replace(return_annotation=inspect.Signature.empty)
args_index = {}
for i, arg in enumerate(signature.parameters.keys()):
args_index[arg] = (i, arg)
def print_results(grouped, indent, result_fn):
nonlocal buf
space = False
for arg, results in grouped:
if space:
buf += "\n"
else:
space = True
buf += f"\n{indent}# {arg}:"
for result in results:
buf += f"\n{indent}{result_fn(result)}"
buf = ""
indent = 4 * " "
if self._static_results:
grouped_static_results = group(self._static_results, args_index)
buf += "\nThe following dimensions have been specialized and CANNOT be dynamic."
buf += f"\n```\ndef specializations{str(signature)}:"
print_results(
grouped_static_results,
indent,
lambda result: f"assert {result}",
)
buf += "\n```\n"
if self._dynamic_results:
grouped_dynamic_results = group(self._dynamic_results, args_index)
buf += "\nThe following dimensions CAN be dynamic."
buf += "\nYou can use the following code to specify the constraints they must satisfy:"
buf += f"\n```\ndef specify_constraints{str(signature)}:"
buf += f"\n{indent}return ["
print_results(
grouped_dynamic_results,
indent * 2,
lambda result: f"{remove_default_lower_bound(result)},",
)
buf += f"\n{indent}]\n```\n"
return buf
TLS = threading.local()
class ShapeEnvLoggerAdapter(logging.LoggerAdapter):
def process(self, msg, kwargs):
# TODO: Maybe suppress the envid if not DEBUG?
return '%s: %s' % (self.extra['envid'], msg), kwargs
ENV_COUNTER = collections.Counter()
class ShapeEnv:
def __init__(
self, *,
allow_scalar_outputs=True,
allow_dynamic_output_shape_ops=True,
# NB: These are legacy configuration that help us make good choices
# when the constraint/dynamic dims are not explicitly passed to us.
# Ideally we will fix all call sites to be explicit and not have
# implicit choices, but this apparently was pretty involved.
assume_static_by_default=False,
# Note - On 0/1 specialization
#
# The following options affect decisions we make about eager
# specialization. Disabling them will increase trace time (as we do
# more symbolic reasoning) and can also harm the quality of generated
# code (because inductor may not be able to specialize for bounds
# being equal--although if we later respecialize because of a guard,
# your code may be just as good as it was before.)
#
# When True, eagerly specialize input sizes which have 0/1.
specialize_zero_one=True,
# When True, assume input sizes which have the same size are
# symbolically equal.
duck_shape=True,
# For debugging
frame_id=None,
co_fields=None,
):
# Not directly used by ShapeEnv; indirectly used by FakeTensor
self.allow_scalar_outputs = allow_scalar_outputs
self.allow_dynamic_output_shape_ops = allow_dynamic_output_shape_ops
self.guards: List[ShapeGuard] = []
# Maps symbolic ints to their original concrete values
# Currently populated from tensors
self.var_to_val: Dict["sympy.Symbol", "sympy.Integer"] = {}
# Maps symbolic ints to their min/max range. These ranges
# are conservative: the int MUST fall in the range, but the
# range may contain ints which may not actually appear in
# practice
self.var_to_range: Dict["sympy.Symbol", ValueRanges] = {}
self.var_to_sources: Dict["sympy.Symbol", List[Source]] = {}
self.var_to_stack: Dict["sympy.Symbol", traceback.StackSummary] = {}
# Maps from sympy ints to expressions representing them
# Populated from equality guards (i.e. a.shape[0] == b.shape[0])
self.replacements: Dict["sympy.Symbol", "sympy.Expr"] = {} #
# Set holds a % b expressions that evaluate to 0.
self.divisible: Set["sympy.Expr"] = set()
# Duck-shaping says that if two input tensors have the same size,
# they get assigned the same symbolic variable
self.val_to_var: Dict[int, "sympy.Expr"] = {}
if specialize_zero_one:
self.val_to_var = {0: sympy.Integer(0), 1: sympy.Integer(1)}
self.unbacked_symfloat_counter = itertools.count()
self.unbacked_symint_counter = itertools.count()
self.assume_static_by_default = assume_static_by_default
self.specialize_zero_one = specialize_zero_one
self.duck_shape = duck_shape
per_frame_id = ENV_COUNTER[frame_id]
ENV_COUNTER[frame_id] += 1
if frame_id is None:
env_id = per_frame_id
else:
env_id = f"{frame_id}.{per_frame_id}"
self.log = ShapeEnvLoggerAdapter(log, {'envid': env_id})
self.log.info("create_env")
self.frozen = False
self.dim_constraints: Optional[DimConstraints] = None
self.counter = collections.Counter()
# A selection of important fields on co_field; solely used for
# signpost_event
self.co_fields = co_fields if co_fields else {}
def freeze(self):
self.frozen = True
def _suppress_guards_tls(self):
return getattr(TLS, "suppress_guards", False)
@contextmanager
def suppress_guards(self):
TLS.suppress_guards = True
try:
yield
finally:
TLS.suppress_guards = False
def _get_key(self):
"""
Defines the current "state" of the guards we've accumulated in this ShapeEnv.
Determines when we need to invalidate our cache
"""
return (len(self.replacements), len(self.divisible))
def _produce_dyn_sizes(self,
ex: torch.Tensor,
source: Source,
dynamic_dims: DimList[DimDynamic],
constraint_dims: DimList[DimConstraint],
) -> List[sympy.Expr]:
from torch._dynamo.source import TensorPropertySource, TensorProperty
size = []
for i, val in enumerate(ex.size()):
size.append(self.create_symbol(
val, TensorPropertySource(source, TensorProperty.SIZE, i), dynamic_dims[i], constraint_dims[i]
))
return size
def create_symbolic_sizes_strides_storage_offset(
self,
ex: torch.Tensor,
source: Source,
*,
dynamic_dims: Optional[DimList[DimDynamic]] = None,
constraint_dims: Optional[DimList[DimConstraint]] = None,
):
"""
Returns a list of symbolic sizes and strides for the given tensor.
We try our best to express stride in terms of the sizes, so as to not
introduce new symbolic variables.
"""
dim = ex.dim()
# Reimplement the legacy behavior
if constraint_dims is None:
constraint_dims = [None] * dim
if dynamic_dims is None:
dynamic_dims = []
for i in range(dim):
# NB: This is encapsulation breaking! Legacy behavior was
# bad.
if _is_dim_dynamic(ex, i):
r = DimDynamic.DYNAMIC
elif self.assume_static_by_default:
r = DimDynamic.STATIC
else:
r = DimDynamic.DUCK
dynamic_dims.append(r)
dynamic_dims = [DimDynamic.DUCK] * dim
# TODO: make this configurable from outside policy; we made a policy
# decision here where if all sizes are static, we are going to
# specialize all of the inner strides/offset too. We don't have to
# do this, and arguably we should ALWAYS allow for dynamic offset,
# this is cheap.
# TODO: This should be DYNAMIC, using DUCK for BC
dynamic_strides_offset = DimDynamic.STATIC if all(r == DimDynamic.STATIC for r in dynamic_dims) else DimDynamic.DUCK
assert len(dynamic_dims) == dim
assert len(constraint_dims) == dim
from torch._dynamo.source import TensorPropertySource, TensorProperty
size: List[sympy.Expr] = self._produce_dyn_sizes(ex, source, dynamic_dims, constraint_dims)
stride: List[Optional[sympy.Expr]] = [None] * len(size)
for i, val in enumerate(ex.stride()):
if val in (0, 1):
stride[i] = sympy.Integer(val)
while any(x is None for x in stride):
candidates = {
ex.size(i) * ex.stride()[i]: size[i] * stride[i]
for i in range(len(size))
if stride[i] is not None and ex.stride()[i] >= 0
}
# iterate over unbound strides in sorted order
val_list = sorted(
[(ex.stride()[i], i) for i in range(len(stride)) if stride[i] is None]
)
for _, i in val_list:
if stride[i] is None and ex.stride()[i] in candidates:
stride[i] = candidates[ex.stride()[i]]
candidates[ex.size(i) * ex.stride()[i]] = size[i] * stride[i]
if any(x is None for x in stride):
# bind the smallest unbound stride to a new variable
val, i = min(
[
(ex.stride()[i], i)
for i in range(len(stride))
if stride[i] is None
]
)
stride[i] = self.create_symbol(
val,
TensorPropertySource(source, TensorProperty.STRIDE, i),
dynamic_dim=dynamic_strides_offset,
constraint_dim=None,
)
assert all(x is not None for x in stride)
sym_sizes = [self.create_symintnode(i, hint=hint) for i, hint in zip(size, ex.size())]
sym_stride = []
for i, stride_expr in enumerate(stride):
# NB: Don't duck size the stride; instead use the expression
# we computed
assert stride_expr is not None
sym_stride.append(self.create_symintnode(stride_expr, hint=ex.stride(i)))
sym_storage_offset = self.create_symintnode(self.create_symbol(
ex.storage_offset(),
TensorPropertySource(source, TensorProperty.STORAGE_OFFSET),
dynamic_dim=dynamic_strides_offset,
constraint_dim=None,
), hint=ex.storage_offset())
return sym_sizes, sym_stride, sym_storage_offset
# If you know what the current hint value of the SymInt to be created
# is, pass it into hint. Otherwise, pass None and we will make our best
# guess
def create_symintnode(self, sym: "sympy.Expr", *, hint: Optional[int]):
if isinstance(sym, sympy.Integer):
if hint is not None:
assert int(sym) == hint
return int(sym)
return SymInt(SymNode(sym, self, int, hint))
def create_symboolnode(self, sym: "sympy.Expr"):
return SymBool(SymNode(sym, self, bool, None))
def create_unbacked_symfloat(self):
symbol = sympy.Symbol(f"f{next(self.unbacked_symfloat_counter)}")
self.counter["create_unbacked_symbol"] += 1
self.var_to_stack[symbol] = traceback.extract_stack()[:-1]
self.var_to_range[symbol] = ValueRanges.unknown()
return SymFloat(SymNode(symbol, self, float, None))
def create_unbacked_symint(self):
symbol = sympy.Symbol(f"i{next(self.unbacked_symint_counter)}", integer=True)
self.counter["create_unbacked_symbol"] += 1
self.var_to_stack[symbol] = traceback.extract_stack()[:-1]
self.var_to_range[symbol] = ValueRanges(-sys.maxsize - 1, sys.maxsize)
return self.create_symintnode(symbol, hint=None)
def create_unbacked_symbool(self):
symbol = sympy.Symbol(f"i{next(self.unbacked_symint_counter)}", integer=True)
self.counter["create_unbacked_symbol"] += 1
self.var_to_stack[symbol] = traceback.extract_stack()[:-1]
self.var_to_range[symbol] = ValueRanges(0, 1)
return self.create_symboolnode(sympy.Eq(symbol, 1))
def create_symbol(
self,
val: int,
source: Source,
dynamic_dim: DimDynamic = DimDynamic.DUCK,
constraint_dim: DimConstraint = None, # NB: includes None
) -> "sympy.Expr":
assert isinstance(source, Source), f"{type(source)} {source}"
# It's always sound to allocate a symbol as DYNAMIC. If the user
# constrained the symbol, force the policy to DYNAMIC, because our
# constraint code will do weird stuff if, e.g., it's duck shaped
if constraint_dim is not None:
dynamic_dim = DimDynamic.DYNAMIC
if dynamic_dim is DimDynamic.STATIC:
return sympy.Integer(val)
elif dynamic_dim is DimDynamic.DUCK:
# duck_shape can be used to globally turn off duck shaping, even
# if it was requested
duck = self.duck_shape
elif dynamic_dim is DimDynamic.DYNAMIC:
duck = False
else:
raise AssertionError(f"unhandled dynamic_dim {dynamic_dim}")
if val < 0:
from torch._dynamo.source import NegateSource
assert constraint_dim is None, "constraints on negative unspec ints NYI"
return -self.create_symbol(-val, NegateSource(source), dynamic_dim, constraint_dim)
if val in (0, 1) and self.specialize_zero_one:
r = self.val_to_var[val]
elif not duck or val not in self.val_to_var:
# If we're not duck shaping, we always create a new symbol
# Even if we're duck shaping, if we haven't seen this particular
# value before, we also create a new symbol
sympy_expr = sympy.Symbol(f"s{len(self.var_to_val)}", positive=True, integer=True)
self.log.info("create_symbol %s = %s for %s", sympy_expr, val, source.name())
self.counter["create_symbol"] += 1
# We always associate vars to vals
self.var_to_val[sympy_expr] = sympy.Integer(val)
# Do the appending later, because we always want to populate this
self.var_to_sources[sympy_expr] = []
if duck:
# Make sure to reuse this symbol for subsequent duck shaping
self.val_to_var[val] = sympy_expr
# Apply default range, which assumes not zero-one
self.var_to_range[sympy_expr] = self._default_value_range()
# Small performance optimization: if we have a min-max constraint,
# we can proactively narrow to that range
if isinstance(constraint_dim, StrictMinMaxConstraint):
assert not duck
self.var_to_range[sympy_expr] &= constraint_dim.vr
vr = self.var_to_range[sympy_expr]
if val not in vr:
raise ConstraintViolationError(f"{val} not in range [{vr.lower}, {vr.upper}]")
r = sympy_expr
else:
# This implements duck-shaping: input sizes that match are assigned
# the same symint
r = self.val_to_var[val]
self.log.debug("create_symbol %s duck sized %s", r, source.name())
if isinstance(r, sympy.Symbol):
self.var_to_sources[r].append(source)
return r
# Generates a list of guards strings which, when evaluated in a context that
# defines tensors for all the sources, returns True or False depending
# on if the guards in the list evaluated to True or not. Primarily used by Dynamo,
# but this is also helpful for manual testing of guards (see
# evaluate_guards_for_args)
#
# For convenience in testing, a source is allowed to be a str,
# in which case we will assume it is a LocalSource
#
# simplified lets you omit duck sizing, equality and 0/1 guards.
# This is useful for testing when you don't care about the boilerplate
# guards, and it may be helpful for user output too (be careful though;
# some equality guards are nontrivial! It would be nice to get simplified
# output to print them too). It's private because it's not
# intended for normal use
def produce_guards(
self,
placeholders,
sources,
source_ref=lambda n: n.name(),
*,
# An input is either a SymInt (in which case you directly have
# DimConstraint) or a Tensor (in which case you have a
# DimList[DimConstraint]). Whenever Optional is accepted, that
# just means there are no constraints
constraint_inputs: Optional[InputList[Union[DimConstraint, Optional[DimList[DimConstraint]]]]] = None,
equalities_inputs: Optional[Set[Tuple[Source, Source]]] = None,
_simplified=False,
# Indicates if we should produce guards for known static values.
ignore_static=True,
) -> List[str]:
self.log.info("produce_guards")
assert len(placeholders) == len(sources)
# Expand optional inputs, or verify invariants are upheld
if constraint_inputs is None:
constraint_inputs = [
[None] * t.dim() if isinstance(t, torch.Tensor) else None for t in placeholders
]
else:
assert len(constraint_inputs) == len(placeholders)
for i, (t, constraint) in enumerate(zip(placeholders, constraint_inputs)):
if isinstance(t, torch.Tensor):
if constraint is None:
constraint_inputs[i] = [None] * t.dim()
else:
assert len(constraint) == t.dim()
else:
assert isinstance(t, (SymInt, int))
assert not isinstance(constraint, list)
# It took a lot of sweat to figure out the algorithm here. Let's
# explain how it works.
#
# The ShapeEnv lifecycle looks something like this:
#
# - For each input, you either generate a fresh Sympy symbol (s0) to
# represent its value (a binding site), or you reuse some
# preexisting symbol or expression, skipping the symbol allocation
# (e.g., duck sizing to a preexisting symbol, or expressing a
# stride as a multiplication of a separate stride and size.)
# Naively, you might expect to bind a fresh Sympy symbol for
# every input, but this is fairly wasteful as most of these
# symbols immediately simplify away, and if you don't eagerly
# specialize, e.g., 0/1 symbols, you end up with very complicated
# expressions that are not optimizable in practice.
#
# - You perform some compute on these symbols, occasionally
# introducing guards on boolean expressions on these symbols.
# In particular, whenever we guard on equality (_maybe_guard_eq),
# we can simplify shapes; e.g., when s0 == s1 * 2, we can now
# replace all occurrences of s0 with s1 * 2. Sometimes, a
# boolean expression evaluation doesn't introduce a guard, as
# the guard is already entailed by the simplifications we have
# applied.
#
# - In the end, you have a bunch of replacements (saying how to
# simplify shapes) and a bunch of guards (all the equality guards
# are trivial, because they're covered by the replacements).
#
# From the ShapeEnv, we must generate a Python expression that, when
# evaluated on a set of inputs, tells us whether or not these boolean
# expressions would have evaluated in the same way. However,
# we cannot easily compute this, as we elide recording boolean
# expressions when we think they are vacuously true. Thus, we seek
# an approximation: we must generate an expression, if true, would have
# produced an "equivalent" ShapeEnv, which would answer guard
# expressions in the same way.
#
# Our notion of equivalence is a bit subtle. For example, consider
# the ShapeEnv created from an input of size (5, 4) versus (4, 4)
# (no other guards.) Duck sizing would generate (s0, s1) in the first
# case but (s0, s0) in the second. We do NOT assume that size
# variables are disjoint; so in fact a graph that assumes the input
# could be (s0, s1) subsumes (s0, s0) (setting s0 == s1), but not
# vice versa. However, consider an analogous case (1,) versus (2,).
# Duck sizing generates (1,) and (s0,); the (s0,) graph does NOT
# subsume the (1,) graph because we assume that any size variables
# is NOT 0/1 (and make simplifications according to this; e.g., if
# we queried s0 == 0, we would immediately return False without
# returning a guard.)
#
# So, it is perhaps easier to flip things on their head: the guard
# expressions we generate here say what simplifications are valid,
# and what are not. Below, we explain each of the guard expressions
# we generate
# TODO: Make this more efficient by binding all the size/stride/offsets
# to locals before performing tests on them.
from torch._dynamo.source import TensorPropertySource, TensorProperty, NegateSource
# Actual codegen must be delayed as we don't necessarily know what
# the symbol mapping is
input_guards = []
symbol_to_source = collections.defaultdict(list)
symbol_to_constraints = collections.defaultdict(list)
constraint_violations : List[Tuple[bool, Callable[[], str]]] = []
def record_constraint_violation(warn_only, msg, hint=None):
constraint_violations.append(
(warn_only, lambda: f"{msg} {hint()}" if hint else msg)
)
def is_dim(src):
return isinstance(src, TensorPropertySource) and src.prop is TensorProperty.SIZE
# How do we know what the value of s0 is? Fresh variables can only be
# bound by inputs, so there MUST be some other input which binds the
# variable. If there is no such input, this is an error in our
# system. We record where all symbols come from, to help you diagnose
# why those symbols didn't occur.
#
# In fact, generally speaking it is only possible for the "outermost"
# user of a ShapeEnv to evaluate the guards, because some inputs may
# not be available to inner levels. For example, Dynamo can guard on
# tensors that never actually become graph arguments (they are
# pruned). In this case, only Dynamo knows about these arguments.
def track_symint(source, val, constraint=None):
if isinstance(val, SymInt):
s = val.node.expr
if isinstance(s, sympy.Symbol):
symbol_to_source[s].append(source)
if constraint is not None:
symbol_to_constraints[s].append(constraint)
elif isinstance(-s, sympy.Symbol):
symbol_to_source[-s].append(NegateSource(source))
else:
constraint_violated = False
if isinstance(constraint, StrictMinMaxConstraint):
constraint_violated = True
elif isinstance(constraint, RelaxedUnspecConstraint):
if s.free_symbols:
# TODO: Maybe non-strict constraint shouldn't error
# here? Check what happens in practice
constraint_violated = True
else:
i = int(s)
# Don't complain about 0/1 specialization, we
# expect to have to compile in this case anyway
if i not in (0, 1):
constraint_violated = True
if constraint_violated:
def hint(s):
if s.free_symbols:
return (
f"Perhaps you meant to specify a constraint on {s.free_symbols}?" +
"; ".join(
f"{s0} bound by " + ", ".join(str(source0) for source0 in symbol_to_source[s0])
for s0 in s.free_symbols
)
)
else:
return "Did you really mean to mark this dimension as dynamic?"
msg = (
f"Could not validate constraint {constraint.render(source)} as "
f"{source.name()} is actually a non-atomic symbolic expression "
f"{s}."
)
record_constraint_violation(
constraint.warn_only,
msg,
hint=functools.partial(hint, s),
)
input_guards.append((source, s))
else:
s = sympy.Integer(val)
input_guards.append((source, s))
constraint_violated = False
if isinstance(constraint, StrictMinMaxConstraint):
constraint_violated = True
elif isinstance(constraint, RelaxedUnspecConstraint):
# Don't complain about 0/1 specialization, we
# expect to have to compile in this case anyway
if val not in (0, 1):
constraint_violated = True
if constraint_violated:
msg = (
f"Could not validate constraint {constraint.render(source)} as "
f"{source.name()} was inferred to be constant ({val}). For more information "
"about why it is constant, run with TORCH_LOGS=dynamic"
)
record_constraint_violation(constraint.warn_only, msg)
for t, source, constraint in zip(placeholders, sources, constraint_inputs):
if isinstance(source, str):
from torch._dynamo.source import LocalSource
source = LocalSource(source)
assert isinstance(source, Source)
if t is None:
continue
if isinstance(t, (SymInt, int)):
track_symint(source, t)
continue
assert isinstance(t, torch.Tensor)
for i, ss in enumerate(t.size()):
property_source = TensorPropertySource(source, TensorProperty.SIZE, i)
track_symint(property_source, ss, constraint[i])
for i, ss in enumerate(t.stride()):
track_symint(TensorPropertySource(source, TensorProperty.STRIDE, i), ss)
track_symint(TensorPropertySource(source, TensorProperty.STORAGE_OFFSET), t.storage_offset())
# 1. Every input must equal the final simplified symbolic expression
# stored on the placeholder. Given a placeholder (s0*2, s1),
# if we have an input (2, 3), we must show s0*2 == 2 and s1 == 3.
# This does a lot of work: it covers duck sizing and equality guards.
exprs = []
self.dim_constraints = DimConstraints(
symbol_to_source,
self.var_to_val,
set(symbol_to_constraints.keys()),
)
if not _simplified:
for source, expr in input_guards:
# Small optimization
if (
isinstance(expr, sympy.Symbol) and
symbol_to_source.get(expr) and
source == symbol_to_source[expr][0]
):
continue
# This logic excludes static values found on tensors from guarding, because
# dynamo's check_tensor_fn does that (see guards.cpp).
# However, for non tensor sources, we still need to guard here.
if ignore_static and isinstance(source, TensorPropertySource):
if len(expr.free_symbols) == 0:
self.log.debug("Skipping guard %s", f"{source_ref(source)} == {expr}")
continue
if is_dim(source):
self.dim_constraints.add_equality(source, expr)
sexpr = ShapeGuardPrinter(symbol_to_source, source_ref, self.var_to_sources).doprint(expr)
exprs.append(f"{source_ref(source)} == {sexpr}")
if (
isinstance(expr, sympy.Symbol) and
expr in symbol_to_constraints and
isinstance(source, TensorPropertySource)
and source.prop is TensorProperty.SIZE
and equalities_inputs and
not equalities_inputs.is_equal(source, symbol_to_source[expr][0])
):
msg = (
f"The specified set of equalities {equalities_inputs.render()} "
f"is not sufficient; please also specify {source_ref(source)} == {sexpr}."
)
record_constraint_violation(equalities_inputs.warn_only, msg)
# NB: Not necessary to report constraint violations here:
# constraints are guaranteed to be on symbols (we've already
# caught constants and non-atomic expressions), so we only
# have relational constraints, but we don't support those
# at the moment
# 2. Every guard must evaluate to True (but remember many guards
# like s0 == s1*2 because trivial due to simplification)
for g, tb in self.guards:
if self._maybe_evaluate_static(g) is not None:
continue
g = self.simplify(g)
try:
if any(is_dim(source) for s in g.free_symbols for source in symbol_to_source[s]):
self.dim_constraints.add(g)
guard_expr = ShapeGuardPrinter(symbol_to_source, source_ref, self.var_to_sources).doprint(g)
exprs.append(guard_expr)
# A non-relational constraint on a single sizevar can violate
# a constraint
if len(g.free_symbols) == 1:
symbol = list(g.free_symbols)[0]
source = symbol_to_source[symbol][0]
constraints = symbol_to_constraints[symbol]
for c in constraints:
if isinstance(c, StrictMinMaxConstraint):
msg = (
f"Could not validate (strict) constraint {c.render(source)} as "
f"we generated a guard on this size variable: {guard_expr}."
)
record_constraint_violation(c.warn_only, msg)
elif isinstance(c, RelaxedUnspecConstraint):
# This is fine, we allow guards here as long as it
# didn't constrain it to one value (we don't
# actually know this; this depends on our
# ValueRanges reasoning capability)
pass
else:
raise AssertionError(f"unrecognized constraint {c}")
except Exception:
self.log.warning("Failing guard allocated at: \n%s", tb)
raise
# 3. Every symbol must be within its value range (this handles 0/1
# specialization too). NB: because we never update value ranges
# except in case of explicit user annotation, these are not included
# in simplified. However, when we start updating value ranges
# these should probably get reported in tests too
if not _simplified:
for symbol, sources in symbol_to_source.items():
r = self.var_to_range[symbol]
for c in symbol_to_constraints[symbol]:
if isinstance(c, StrictMinMaxConstraint):
# Refine the user VR based on default value range, as
# no matter what the user specifies, we will have
# narrowed it according to the default range
c_vr = c.vr & self._default_value_range()
# NB: exact match is OK here, because we already
# applied the constraint when we allocated the symbol
# originally. Otherwise, should only assert that
# vr is superset of c_vr
if not (c_vr.lower <= r.lower and c_vr.upper >= r.upper):
msg = (
f"Could not validate constraint {c.render(sources[0])} as "
f"we actually inferred the valid range to be [{r.lower}, {r.upper}]."
)
record_constraint_violation(c.warn_only, msg)
assert sources
assert symbol.is_integer
bounds = []
if r.lower != -sympy.oo:
if any(is_dim(source) for source in sources):
self.dim_constraints.add(sympy.Ge(symbol, r.lower))
bounds.append(str(r.lower))
bounds.append(source_ref(sources[0]))
# NB: This looks like an off-by-one error but it's not: the
# upper bound may be sys.maxsize - 1 because we intentionally
# exclude sys.maxsize from our bounds to deal with direct
# == INT_MAX guards, but it's still dumb to actually test it.
# Note that you can be off by a pretty large constant and it
# won't matter because sizes in practice will be no where near
# the 64-bit limit.
if r.upper != sympy.oo and r.upper < sys.maxsize - 1:
if any(is_dim(source) for source in sources):
self.dim_constraints.add(sympy.Le(symbol, r.upper))
bounds.append(str(r.upper))
if len(bounds) > 1:
exprs.append(" <= ".join(bounds))
if constraint_violations:
warn_msgs = []
error_msgs = []
for warn_only, msg in constraint_violations:
if warn_only:
msg = f" {len(warn_msgs) + 1}. {msg()}"
warn_msgs.append(msg)
else:
msg = f" {len(error_msgs) + 1}. {msg()}"
error_msgs.append(msg)
if len(error_msgs) > 0:
err = '\n'.join(error_msgs)
raise ConstraintViolationError(f"Constraints violated!\n{err}")
elif len(warn_msgs) > 0:
log.debug("%s Warning only constraints violated", len(warn_msgs))
signpost_event(
"dynamic",
"produce_guards",
{
**self.co_fields,
**self.counter,
"num_guards": len(exprs),
"free_symbols": sum(1 for v in symbol_to_source.values() if v),
},
)
return exprs
def evaluate_guards_for_args(self, placeholders, args, *, ignore_static=True):
from torch._dynamo.source import LocalSource
arg_names = [f"t{i}" for i in range(len(args))]
guards = self.produce_guards(placeholders, [LocalSource(a) for a in arg_names], ignore_static=ignore_static)
if guards:
code = " and ".join(guards)
return eval(code, SYMPY_INTERP, {"L": dict(zip(arg_names, args))})
return True
def bind_symbols(self, placeholders, args):
# Given a paired list of placeholders (fake tensors with
# symbolic sizes) and concrete arguments (regular tensors
# with real sizes), returns a dictionary mapping each
# symbol to its real value. So for example, if you
# have a placeholder with size (s0, s1), binding
# (2, 4) to it will give you {s0: 2, s1: 4}. This is
# not guaranteed to bind ALL symbols in the ShapeEnv;
# we can't bind a symbol if it doesn't occur in any placeholder,
# and symbols that already have replacements won't get bindings.
# This is a little duplicative with evaluate_guards but
# it's different enough that it seemed cleanest to make
# another copy. This assumes the guards are already checked,
# though if it's cheap we'll check for shenanigans
bindings: Dict[sympy.Symbol, int] = {}
def bind_symint(arg, val):
if isinstance(val, SymInt):
s = val.node.expr
if isinstance(s, sympy.Symbol):
if s in bindings:
assert bindings[s] == arg, f"{bindings[s]} != {arg}"
else:
bindings[s] = arg
elif isinstance(-s, sympy.Symbol):
if -s in bindings:
assert bindings[-s] == -arg, f"{bindings[-s]} != {-arg}"
else:
bindings[-s] = -arg
for t, arg in zip(placeholders, args):
if t is None:
continue
if isinstance(t, SymInt):
bind_symint(arg, t)
continue
assert isinstance(t, torch.Tensor)
for i, s in enumerate(t.size()):
bind_symint(arg.size(i), s)
for i, s in enumerate(t.stride()):
bind_symint(arg.stride(i), s)
bind_symint(arg.storage_offset(), t.storage_offset())
return bindings
def get_nontrivial_guards(self):
return [self.simplify(guard.expr) for guard in self.guards if self._maybe_evaluate_static(guard.expr) is None]
def format_guards(self, verbose=False):
def format_tb(tb):
if not verbose:
return ""
return f"\n Guarded at:\n{textwrap.indent(tb, ' ')}"
return '\n'.join(f" - {guard.expr}{format_tb(guard.stack)}" for guard in self.guards)
def get_shape_groups(self):
shape_groups = collections.defaultdict(list)
for k, v in self.replacements.items():
shape_groups[v].append(k)
return shape_groups
@_lru_cache
def _maybe_evaluate_static(self, expr: "sympy.Expr", *, unbacked_only: bool = False) -> "Optional[sympy.Expr]":
"""
Tries to evaluate expr without introducing guards
"""
expr = self.simplify(expr)
# Simplify making use of value range lower bound
symbols = list(expr.free_symbols)
new_shape_env = {}
new_range_env = {}
for idx, k in enumerate(symbols):
vr = self.var_to_range[k]
# Don't do anything if we don't have a nontrivial lower bound
# Also don't do anything if we asked only to simplify unbacked
# SymInt
if vr.lower == -sympy.oo or (unbacked_only and k in self.var_to_val):
new_range_env[k] = vr
continue
# Positive means >= 1
# Positive - 1 means >= 0
# Positive + lower - 1 means >= lower
# The new symbol 's' is "too low", so when we substitute it in
# we have to increase it by offset (and conversely, the new
# variables have to have their value range bounds adjusted as
# well)
s = sympy.Symbol(f"shape_{idx}", positive=True, integer=True)
offset = vr.lower - 1
new_shape_env[k] = s + offset
new_range_env[s] = ValueRangeAnalysis.sub(vr, offset)
def replace(expr, repl):
return expr.xreplace(repl)
try:
new_expr = replace(expr, new_shape_env)
except RecursionError:
log.warning("RecursionError in sympy.xreplace(%s, %s)", expr, new_shape_env)
self.counter["sympy_recursion_error"] += 1
return None
floor_div_replace = {}
for atom in new_expr.atoms(FloorDiv):
floor_div_replace[atom] = sympy.floor(atom.args[0] / atom.args[1])
new_expr = safe_expand(new_expr.xreplace(floor_div_replace))
# TODO: when unbacked_only, can sometimes early return even when there
# are still free symbols
if len(list(new_expr.free_symbols)) == 0:
return new_expr
# Check if the range can solve it statically
out = sympy_interp(ValueRangeAnalysis, new_range_env, new_expr)
if out.is_singleton():
return out.lower
return new_expr if unbacked_only else None
@_lru_cache
def replace(self, expr: "sympy.Expr") -> "sympy.Expr":
replacements = {s: self._find(cast(sympy.Symbol, s)) for s in expr.free_symbols}
return safe_expand(expr.xreplace(replacements))
@_lru_cache
def _update_divisible(self):
new_divisible = set()
for k in self.divisible:
res = self.replace(k)
if len(res.free_symbols) > 0:
new_divisible.add(k)
self.divisible = new_divisible
@_lru_cache
def simplify(self, expr: "sympy.Expr") -> "sympy.Expr":
expr = self.replace(expr)
# TODO it would seem that this pass is not necessary given the
# below replacement of // with /, but for nested FloorDivs
# the non-recursive replacement doesn't work, and
# recursive makes it hard to look up divisibility,
# because existing divisibility info has FloorDiv in it, not /
# for now just do a separate pass to catch common nested case
if expr.has(FloorDiv):
self._update_divisible()
div_replacements = {}
for atom in expr.atoms(FloorDiv):
base, divisor = atom.args
if isinstance(divisor, FloorDiv):
base1, divisor1 = divisor.args
if self.replace(base % divisor) in self.divisible and \
base == base1 and self.replace(base1 % divisor1) in self.divisible:
div_replacements[atom] = divisor1
expr = expr.xreplace(div_replacements)
expr = safe_expand(expr)
if expr.has(FloorDiv):
div_replacements = {}
pows = expr.atoms(sympy.Pow)
rationals = expr.atoms(sympy.Rational).difference(expr.atoms(sympy.Integer))
for fd in expr.atoms(FloorDiv):
base, divisor = fd.args
if self.replace(base % divisor) in self.divisible:
div_replacements[fd] = base / divisor
new_expr = expr.xreplace(div_replacements)
new_expr = safe_expand(new_expr)
new_pows = new_expr.atoms(sympy.Pow)
new_rationals = new_expr.atoms(sympy.Rational).difference(new_expr.atoms(sympy.Integer))
# divisions simplified away
if new_pows.issubset(pows) and new_rationals.issubset(rationals):
expr = new_expr
return expr
@lru_cache(256)
def size_hint(self, expr: "sympy.Expr"):
"""
Gets a size hint for a given expression from the underlying shapes we had.
Does not introduce a guard, so only use this when you can guarantee that
your code is still valid for arbitrary shapes (such as optimization decisions)
"""
result_expr = safe_expand(expr).xreplace(self.var_to_val)
if len(result_expr.free_symbols) != 0:
r = self._maybe_evaluate_static(result_expr)
if r is not None:
return r
raise self._make_data_dependent_error(result_expr, expr)
return result_expr
def _make_data_dependent_error(self, expr, unhinted_expr):
# TODO: in a Dynamo context, having user code, and having the
# name of the local, will be much better
for s in expr.free_symbols:
stacktrace = ''.join(traceback.format_list(self.var_to_stack[s]))
self.log.debug("Data dependent variable '%s' allocated at:\n%s", s, stacktrace)
return GuardOnDataDependentSymNode(
"It appears that you're trying to get a value out of symbolic int/float "
"whose value is data-dependent (and thus we do not know the true value.) "
f"The expression we were trying to evaluate is {expr} (unhinted: {unhinted_expr}). "
"Scroll up to see where each of these data-dependent accesses originally occurred."
# TODO: Help text about how to use our runtime tests to fix this
# problem
)
def _set_replacement(self, a: "sympy.Symbol", expr: "sympy.Expr") -> None:
"""
Adds or updates a replacement for a symbol.
Use this instead of `self.replacements[a] = expr`.
"""
if torch._dynamo.config.print_specializations and isinstance(expr, (sympy.Integer, sympy.Float)):
# specializing to a constant, which is likely unexpected
# NOTE(avik): It is possible that we try logging the same specialization multiple times, e.g.,
# when adding a to self.replacements, and again when simplifying an expression containing a.
# Thus to avoid duplication, checking whether a is in self.replacements isn't enough; if it is,
# it must not already map to `expr`. Fortunately this check is cheap because `expr` is a constant.
if a not in self.replacements or expr != self.replacements[a]:
self.log.warning("Specializing %s to %s", self.var_to_sources[a][0].name(), expr)
self.log.debug("SPECIALIZATION", stack_info=True)
self.replacements[a] = expr
@_lru_cache
def _find(self, a: "sympy.Symbol") -> "sympy.Expr":
"""
Implements a DSU-like algorithm to find the variable that represents a
Also handles transitive non-identity replacements.
a: b + c
c: d
"""
if a not in self.replacements:
return a
res = self.replacements[a]
cur_replace = {s: self._find(s) for s in res.free_symbols}
self._set_replacement(a, self.replacements[a].xreplace(cur_replace))
return self.replacements[a]
@lru_cache(256)
def _maybe_guard_eq(self, expr: Union["sympy.Eq", "sympy.Ne"], concrete_bool: bool) -> None:
"""
Evaluates the result of an eq call. If true, uses information to
simplify shapes (i.e. a == b or a % 5 == 0)
"""
assert type(concrete_bool) is bool
if isinstance(expr, sympy.Eq):
if not concrete_bool:
return
# NB: Apparently this is load bearing; to see what test fails if
# you comment it out run:
# python test/functorch/test_aotdispatch.py -k
# test_aot_autograd_symbolic_module_exhaustive_nn_LazyConv3d_cpu_float32
elif isinstance(expr, sympy.Ne):
if concrete_bool:
return
free = list(expr.free_symbols)
assert len(free) > 0, f"The expression should not be static by this point: {expr}"
# In case of really gnarly expression, we don't blow up
if len(free) > 5:
return
free = sorted(free, key=lambda x: (self.size_hint(x), x.name), reverse=True) # type: ignore[attr-defined]
lhs = expr.lhs
rhs = expr.rhs
if not expr.has(sympy.Mod):
try:
floor_div_atoms = lhs.atoms(FloorDiv).union(rhs.atoms(FloorDiv))
if len(floor_div_atoms) > 0 and any(a.divisor != 1 for a in floor_div_atoms):
raise NotImplementedError
solutions = sympy.solve(lhs - rhs, free[0], dict=True)
if len(solutions) != 1:
return
solution = solutions[0][free[0]]
if all(t.is_integer for t in sympy.preorder_traversal(solution)):
new_var = self._find(solution)
self._set_replacement(cast(sympy.Symbol, free[0]), new_var)
except NotImplementedError:
pass
except RecursionError:
self.counter["sympy_recursion_error"] += 1
self.log.warning("RecursionError in sympy.solve(%s - %s, %s)", lhs, rhs, free[0])
if expr.has(sympy.Mod):
mod_expr = tuple(expr.atoms(sympy.Mod))[0]
try:
solutions = sympy.solve(lhs - rhs, mod_expr, dict=True)
if len(solutions) == 1 and solutions[0][mod_expr] == 0:
self.divisible.add(mod_expr)
except NotImplementedError:
pass
return
# See: Note - On 0/1 specialization
# NB: sys.maxsize is NOT allowed for sizes, because we use MAX_INT
# as a sentinel sometimes. Your sizevar isn't going to be
# anywhere near the max 64-bit integer anyway.
def _default_value_range(self) -> ValueRanges:
lower = 2 if self.specialize_zero_one else 0
return ValueRanges(lower, sys.maxsize - 1)
@_lru_cache
def _simplify_floor_div(self, expr):
floor_divs = tuple(expr.atoms(FloorDiv))
# we expect floor_divs to be exact,
# and thus add the guards for the exact floordivs,
# even if tracing doesn't require them otherwise
for fd in reversed(floor_divs):
base, divisor = fd.args
mod_expr = sympy.Mod(base, divisor)
eq_expr = sympy.Eq(mod_expr, 0)
# add necessary mod guards
self.evaluate_expr(eq_expr)
return self.simplify(expr)
@lru_cache(256)
def evaluate_expr(self, orig_expr: "sympy.Expr", hint=None):
"""
Given an expression, evaluates it, adding guards if necessary
"""
if len(orig_expr.free_symbols) == 0:
self.log.debug("eval %s [trivial]", orig_expr)
return orig_expr
expr = orig_expr
static_expr = self._maybe_evaluate_static(expr)
if static_expr is not None:
self.log.debug("eval %s == %s [statically known]", orig_expr, static_expr)
return static_expr
if not (expr.free_symbols <= self.var_to_val.keys()):
# TODO: dedupe this with _maybe_evaluate_static
# Attempt to eliminate the unbacked SymInt
new_expr = self._maybe_evaluate_static(expr, unbacked_only=True)
if not (new_expr.free_symbols <= self.var_to_val.keys()):
raise self._make_data_dependent_error(expr.xreplace(self.var_to_val), expr)
expr = new_expr
if hint is None:
concrete_val = self.size_hint(expr)
else:
concrete_val = sympy.sympify(hint)
if self.frozen:
self.counter["ignored_backward_guard"] += 1
signpost_event(
"dynamic",
"evaluate_expr_frozen",
{
**self.co_fields,
"ignored_guard": f"{expr} == {concrete_val}",
},
)
log.warning("Ignored guard %s == %s, this could result in accuracy problems", expr, concrete_val)
if isinstance(expr, (sympy.Eq, sympy.Ne)):
self._maybe_guard_eq(expr, bool(concrete_val))
# TODO: If we successfully eliminate a symbol via equality, it
# is not actually necessary to save a guard for the equality,
# as we will implicitly generate a guard when we match that
# input against the symbol
elif isinstance(concrete_val, sympy.Integer):
# WARNING: we cannot actually do simplifications on guards
# on floating point values, because Sympy generally does not
# think expressions on integers can ever be equal to floating
# point (e.g., sympy.Eq(s0/6, 0.5) evaluates to False). Without
# very clear algebraic laws that hold for floating point, such
# simplifications are error prone anyway, so be sure not to
# maybe_guard_eq in those cases.
self._maybe_guard_eq(sympy.Eq(expr, concrete_val), True)
if concrete_val is sympy.true:
g = expr
elif concrete_val is sympy.false:
g = sympy.Not(expr)
else:
g = sympy.Eq(expr, concrete_val) # type: ignore[arg-type]
if not self._suppress_guards_tls():
tb = traceback.extract_stack()[:-1]
stack = ''.join(traceback.format_list(tb))
guard = ShapeGuard(g, stack)
self.guards.append(guard)
if self.log.isEnabledFor(logging.INFO):
for frame in reversed(tb):
if frame.filename not in uninteresting_files():
break
# NB: this stack is truncated, but it's fine because the main
# stack_info will give you the rest of the info you need
maybe_user_loc = ""
user_tb = TracingContext.extract_stack()
if user_tb:
maybe_user_loc = " at " + format_frame(user_tb[-1])
is_debug = self.log.isEnabledFor(logging.DEBUG)
maybe_extra_debug = ""
if is_debug and user_tb:
maybe_extra_debug = (
'\nUser Stack (most recent call last):\n' +
' (snipped, see stack below for prefix)\n' +
''.join(traceback.format_list(user_tb))
)
self.log.info(
"eval %s [guard added]%s (%s)%s",
g,
maybe_user_loc,
format_frame(frame),
maybe_extra_debug,
stack_info=is_debug,
)
else:
self.log.debug("eval %s [guard suppressed]", g)
return concrete_val
def _is_int(expr):
if not isinstance(expr, SymInt):
return False
if len(expr.node.expr.free_symbols) > 0:
return False
return True
# WARNING: This is legacy, DO NOT USE
def _is_dim_dynamic(t, d):
return hasattr(t, "_dynamo_dynamic_indices") and d in t._dynamo_dynamic_indices