# mypy: allow-untyped-decorators # mypy: allow-untyped-defs import functools import itertools import logging import operator from collections import Counter, defaultdict from typing import Any, Dict, List, Optional, Set, TYPE_CHECKING, Union import torch import torch._inductor as inductor import torch.utils._pytree as pytree from torch import fx from torch._decomp import register_decomposition from torch._dynamo.utils import counters, optimus_scuba_log from torch._inductor import comms from torch._inductor.virtualized import ops from torch._prims_common import is_boolean_dtype, is_expandable_to, is_integer_dtype from torch._utils_internal import upload_graph from torch.fx.experimental.symbolic_shapes import statically_known_true, sym_eq from torch.fx.passes.graph_transform_observer import GraphTransformObserver from .. import config, ir, pattern_matcher from ..codegen.common import BackendFeature, has_backend_feature from ..fx_utils import FakeTensorUpdater, get_fake_args_kwargs, get_node_storage from ..lowering import lowerings as L from ..pattern_matcher import ( _return_true, Arg, CallFunction, CallFunctionVarArgs, filter_nodes, get_arg_value, get_mutation_region_id, Ignored, init_once_fakemode, KeywordArg, ListOf, Match, MULTIPLE, PatternMatcherPass, register_graph_pattern, stable_topological_sort, ) from ..utils import decode_device, get_gpu_type, is_pointwise_use from ..virtualized import V from .b2b_gemm import B2B_GEMM_PASS from .ddp_fusion import fuse_ddp_communication from .group_batch_fusion import group_batch_fusion_passes, POST_GRAD_FUSIONS from .micro_pipeline_tp import micro_pipeline_tp_pass from .pre_grad import is_same_dict, save_inductor_dict from .reinplace import reinplace_inplaceable_ops from .split_cat import POST_GRAD_PATTERNS if TYPE_CHECKING: from sympy import Expr log = logging.getLogger(__name__) aten = torch.ops.aten prims = torch.ops.prims # First pass_patterns[0] are applied, then [1], then [2] pass_patterns = [ PatternMatcherPass(), PatternMatcherPass(), PatternMatcherPass(), ] def post_grad_passes(gm: torch.fx.GraphModule, is_inference: bool): """ Passes that run on after grad. This is called once on the forwards graph and once on the backwards graph. The IR here has been normalized and functionalized. """ if config.dce: # has some issues with mutation in inference mode gm.graph.eliminate_dead_code() if is_inference and config.reorder_for_locality: reorder_for_locality(gm.graph) fake_tensor_updater = FakeTensorUpdater(gm.graph) if config.post_grad_custom_pre_pass is not None: with GraphTransformObserver( gm, "post_grad_custom_pre_pass", config.trace.log_url_for_graph_xform ): config.post_grad_custom_pre_pass(gm.graph) if config.pattern_matcher: lazy_init() optimus_scuba_log["before_recompile_post_grad"] = upload_graph(gm.graph) group_batch_fusion_passes(gm.graph, pre_grad=False) remove_noop_ops(gm.graph) for patterns in pass_patterns: patterns.apply(gm.graph) # type: ignore[arg-type] for pass_name in config.post_grad_fusion_options: # skip all patterns for group batch fusions if pass_name in POST_GRAD_FUSIONS: continue pattern_matcher_pass = POST_GRAD_PATTERNS[pass_name] inductor_before_change = save_inductor_dict( [pattern_matcher_pass.pass_name] ) pattern_matcher_pass.apply(gm.graph) # type: ignore[arg-type] if not is_same_dict(counters["inductor"], inductor_before_change): optimus_scuba_log[ f"{pattern_matcher_pass.pass_name}_post_grad" ] = upload_graph(gm.graph) if config.b2b_gemm_pass: B2B_GEMM_PASS.apply(gm.graph) # type: ignore[arg-type] if config._micro_pipeline_tp: micro_pipeline_tp_pass(gm.graph) if config._fuse_ddp_communication: fuse_ddp_communication( gm.graph, config._fuse_ddp_communication_passes, config._fuse_ddp_bucket_size, ) if config.post_grad_custom_post_pass is not None: with GraphTransformObserver( gm, "post_grad_custom_post_pass", config.trace.log_url_for_graph_xform ): config.post_grad_custom_post_pass(gm.graph) stable_topological_sort(gm.graph) move_constructors_to_gpu(gm.graph) fake_tensor_updater.incremental_update() # Keep these last, since they introduces mutation. Look at # ./fx_passes/README.md for a discussion of mutation invariants. reinplace_inplaceable_ops(gm.graph) decompose_auto_functionalized(gm.graph) comms.reinplace_fsdp_all_gather(gm.graph) gm.recompile() optimus_scuba_log["after_recompile_post_grad"] = upload_graph(gm.graph) gm.graph.lint() @init_once_fakemode def lazy_init(): if torch._C._has_mkldnn: from . import decompose_mem_bound_mm # noqa: F401 from .mkldnn_fusion import _mkldnn_fusion_init _mkldnn_fusion_init() def reorder_for_locality(graph: torch.fx.Graph): def visit(other_node): if ( other_node.op == "call_function" and other_node.target != operator.getitem and all((n in seen_nodes) for n in other_node.users) and get_mutation_region_id(graph, node) == get_mutation_region_id(graph, other_node) ): # move node's producers right before it node.prepend(other_node) seen_nodes = set() # only reorder nodes before the first copy_ in the graph. # copy_ will appear at the end of functionalized graphs when there is mutation on inputs, # and this reordering doesnt work well with mutation first_copy = next( iter(graph.find_nodes(op="call_function", target=torch.ops.aten.copy_.default)), None, ) past_mutating_epilogue = True if first_copy is None else False for node in reversed(graph.nodes): seen_nodes.add(node) if not past_mutating_epilogue: past_mutating_epilogue = node is first_copy continue torch.fx.map_arg((node.args, node.kwargs), visit) def register_lowering_pattern(pattern, extra_check=_return_true, pass_number=1): """ Register an aten to inductor IR replacement pattern """ return pattern_matcher.register_lowering_pattern( pattern, extra_check, pass_dict=pass_patterns[pass_number] ) ################################################################################ # Actual patterns below this point. # Priority of patterns is: # - later output nodes first # - order patterns are defined in ################################################################################ def is_valid_mm_plus_mm(match: Match): *b1, m1, k1 = match.kwargs["mat1"].meta.get("tensor_meta").shape *b2, k2, n1 = match.kwargs["mat2"].meta.get("tensor_meta").shape if k1 != k2: return False *b1, m2, k3 = match.kwargs["mat3"].meta.get("tensor_meta").shape *b2, k4, n2 = match.kwargs["mat4"].meta.get("tensor_meta").shape if k3 != k4: return False if m1 != m2 or n1 != n2: return False return True def scatter_upon_const_tensor_extra_check(m): if not config.optimize_scatter_upon_const_tensor: return False full_shape = m.kwargs["shape"] selector = m.kwargs["selector"] dim = m.kwargs["dim"] if dim < 0: dim += len(full_shape) selector_ft = selector.meta["val"] assert selector_ft.dim() == len(full_shape) for idx, select_sz, full_sz in zip( itertools.count(), selector_ft.shape, full_shape ): if idx == dim: continue # TODO: the pattern can be updated to support the case that index tensor # is shorter. But that will need a more complex condition expression # especially for multi-dimensional tensors. # Skip it for now. if isinstance(full_sz, fx.Node): full_sz = full_sz.meta["val"] if select_sz < full_sz: return False # Actually we can support small size larger than 1. It would be a bit # tedius. E.g., we load all the index values (not many) and compare # them with the position in tensor to decide what value to return. return selector_ft.size(dim) == 1 @register_lowering_pattern( CallFunction( aten.scatter.value, CallFunction( aten.full, KeywordArg("shape"), KeywordArg("background_val"), dtype=KeywordArg("dtype"), ), KeywordArg("dim"), KeywordArg("selector"), KeywordArg("val"), # scalar value ), extra_check=scatter_upon_const_tensor_extra_check, ) def scatter_upon_const_tensor( match: Match, shape, background_val, dtype, dim, selector, val ): """ Match the pattern of full+scatter into a pointwise. TODO: Right now the scatter value must be a scalar. But we could support it when it is a tensor as well. """ from torch._inductor import metrics metrics.num_matches_for_scatter_upon_const_tensor += 1 selector_loader = selector.make_loader() def inner_fn(idx): selector_idx = list(idx) selector_idx[dim] = 0 selector = selector_loader(selector_idx) return ops.where( selector == ops.index_expr(idx[dim], torch.int64), ops.constant(val, dtype), ops.constant(background_val, dtype), ) return ir.Pointwise.create( device=selector.get_device(), dtype=dtype, inner_fn=inner_fn, ranges=shape, ) @register_lowering_pattern( CallFunction( aten.add, CallFunction(aten.mm, KeywordArg("mat1"), KeywordArg("mat2")), CallFunction(aten.mm, KeywordArg("mat3"), KeywordArg("mat4")), ), extra_check=is_valid_mm_plus_mm, ) def mm_plus_mm(match: Match, mat1, mat2, mat3, mat4): return inductor.kernel.mm_plus_mm.tuned_mm_plus_mm(mat1, mat2, mat3, mat4) def cuda_and_enabled_mixed_mm(match): return ( (config.use_mixed_mm or config.mixed_mm_choice != "default") and getattr(match.kwargs["mat1"].meta.get("val"), "is_cuda", False) and ( match.kwargs["mat2_dtype"].itemsize > match.kwargs["mat2"].meta.get("val").dtype.itemsize ) and has_backend_feature("cuda", BackendFeature.TRITON_TEMPLATES) ) def cuda_and_enabled_mixed_mm_and_not_int8(match): return ( cuda_and_enabled_mixed_mm(match) and getattr(match.kwargs["mat1"].meta.get("val"), "is_cuda", False) and getattr(match.kwargs["mat2"].meta.get("val"), "dtype", torch.int8) != torch.int8 ) # bitshift numerics in triton and pytorch don't match for torch.int8 """ this is intended to be used to unpack a [K,N] int4 tensor from a [K/2, N] uint4x2 tensor (where the int4 and uint4x2 are represented with int8 and uint8 respectively) where every other row of the int4 is packed with the row above it as: uint4x2[k,n] = (8+int4[2*k,n])+(8+int4[2*k+1,n])<<4 unpack formulas: int4[2*k,n]=(uint4x2[k,n] & 0xF) - 8 int4[2*k+1,n]=(uint4x2[k,n] >> 4) - 8 thus matching on unpack formula: torch.mm(mat1, torch.cat((mat2 & 0xF, mat2>>4),1).reshape(mat2_mm_shape).to(mat2_dtype).sub(8)) note: although the unpack formula in pytorch and the triton kernel is designed for a uint8 mat2, the behavior of the kernel matches the pytorch formula for all dtypes except torch.int8 where the bitwise numerics in triton do not match those in pytorch. """ @register_lowering_pattern( CallFunction( aten.mm.default, KeywordArg("mat1"), CallFunction( aten.sub.Tensor, CallFunction( prims.convert_element_type.default, CallFunction( aten.reshape.default, CallFunction( aten.cat.default, ListOf( CallFunction( aten.bitwise_and.Scalar, KeywordArg("mat2"), 0xF, ), # CallFunction( # aten.__rshift__.Scalar, # KeywordArg("mat2"), # 4, # ), True, ), 1, ), KeywordArg("mat2_mm_shape"), ), KeywordArg("mat2_dtype"), ), 8, ), ), extra_check=cuda_and_enabled_mixed_mm_and_not_int8, ) def uint4x2_mixed_mm(match: Match, mat1, mat2, mat2_mm_shape, mat2_dtype): return inductor.kernel.unpack_mixed_mm.tuned_uint4x2_mixed_mm( mat1, mat2, mat2_mm_shape, mat2_dtype ) """ torch.mm(mat1, mat2.to(mat2_dtype)) """ @register_lowering_pattern( CallFunction( aten.mm, KeywordArg("mat1"), CallFunction( prims.convert_element_type.default, KeywordArg("mat2"), KeywordArg("mat2_dtype"), ), ), extra_check=cuda_and_enabled_mixed_mm, ) def mixed_mm(match: Match, mat1, mat2, mat2_dtype): return inductor.kernel.mm.tuned_mixed_mm(mat1, mat2, mat2_dtype) @register_graph_pattern( CallFunction( aten.cumsum.default, CallFunction( torch.ops.aten.full.default, KeywordArg("shape"), KeywordArg("fill_value"), dtype=KeywordArg("dtype"), layout=Ignored(), device=KeywordArg("device"), pin_memory=False, _users=MULTIPLE, ), KeywordArg("dim"), _users=MULTIPLE, ), pass_dict=pass_patterns[1], ) def pointless_cumsum_replacement(match: Match, shape, fill_value, device, dtype, dim): """Based on a pattern in OPTForCausalLM""" if is_integer_dtype(dtype) or is_boolean_dtype(dtype): # cumsum promotes all integral types to int64 dtype = torch.int64 def repl(*shape): dim_size = shape[dim] idx = torch.arange(1, dim_size + 1, device=device, dtype=dtype) inter_shape = [1] * len(shape) inter_shape[dim] = dim_size return (idx * fill_value).view(inter_shape).expand(shape) # only replace the output node, not all nodes match.nodes = [match.output_node()] match.replace_by_example(repl, list(shape)) def shape_of_mm(a, b): m, _ = a.get_size() _, n = b.get_size() return [m, n] @register_lowering_pattern( CallFunction(aten.cat, ListOf(CallFunction(aten.mm, Arg(), Arg())), Arg()), ) def cat_mm(match, inputs, dim): return cat_tuned_op(match, inputs, dim, op=L[aten.mm], shape_of=shape_of_mm) @register_lowering_pattern( CallFunction( aten.cat, ListOf(CallFunction(aten.addmm, Arg(), Arg(), Arg())), Arg() ), ) def cat_addmm(match, inputs, dim): def shape_of(bias, a, b): m, _ = a.get_size() _, n = b.get_size() return [m, n] return cat_tuned_op(match, inputs, dim, op=L[aten.addmm], shape_of=shape_of) def cat_tuned_op(match, inputs, dim, *, op, shape_of): """ Memory planning to remove cat. We can't use the stock memory planner since autotuning matmuls needs to know the output layout. """ if len(inputs) == 1: return op(*inputs[0]) # TODO(jansel): rewrite this as a bmm? if dim < 0: dim += len(shape_of(*inputs[0])) assert dim in (0, 1) notdim = 1 - dim new_size: Optional[Union[List[Expr], List[int]]] = None offsets_start = [] offsets_end = [] # compute output sizes for i in range(len(inputs)): shape = shape_of(*inputs[i]) if new_size is None: new_size = shape else: new_size[notdim] = V.graph.sizevars.guard_equals( # type: ignore[call-overload] shape[notdim], new_size[notdim] ) new_size[dim] += shape[dim] offsets_start.append(new_size[dim] - shape[dim]) offsets_end.append(new_size[dim]) assert new_size is not None dtype = functools.reduce( torch.promote_types, [x.get_dtype() for x in itertools.chain.from_iterable(inputs)], ) device = inputs[0][0].get_device() kernel = ir.ConcatKernel( name=None, layout=ir.FixedLayout(device, dtype, new_size), inputs=[], ) kernel_tensor = ir.TensorBox.create(kernel) for i in range(len(inputs)): dst = ir.SliceView.create(kernel_tensor, dim, offsets_start[i], offsets_end[i]) src = op(*inputs[i], layout=dst.get_layout()).data.data assert isinstance(src, (ir.ExternKernelOut, ir.TemplateBuffer)) src.layout = ir.NonOwningLayout(dst) kernel.inputs.append(src) kernel.name = V.graph.register_buffer(kernel) kernel.inputs = ir.ConcatKernel.unwrap_storage(kernel.inputs) V.graph.register_operation(kernel) return kernel_tensor _cat_1 = CallFunction(aten.cat, Arg(), 1, _users=2) @register_lowering_pattern( CallFunction( aten.cat, [ _cat_1, CallFunction( aten.slice, _cat_1, 1, 0, KeywordArg("size"), ), ], 1, ) ) def cat_slice_cat(match, cat_input, size, dim=1): """ This is an example of a more complex pattern where cat_1 is used multiple times inside the pattern. We fold 2 calls to cat into one. Matches: cat_1: f32[1024, 4077] = torch.ops.aten.cat.default([add_26, primals_217], 1) slice_1: f32[1024, 4077] = torch.ops.aten.slice.Tensor(cat_1, 0, 0, 9223372036854775807) slice_2: f32[1024, 19] = torch.ops.aten.slice.Tensor(slice_1, 1, 0, 19) cat_2: f32[1024, 4096] = torch.ops.aten.cat.default([cat_1, slice_2], 1) Rewrite to: slice_2 = torch.ops.aten.slice.Tensor(add_26, 1, 0, 19) cat_2 = torch.ops.aten.cat.default([add_26, primals_217, slice2], 1) """ first, *rest = cat_input # Optimization is optional, because we can just not fold the cat # size should be within first.get_size()[dim] such that the optimization is valid. # For negative `end`, we currently fallback to not optimizing. if size >= 0 and V.graph.sizevars.statically_known_leq(size, first.get_size()[dim]): # fold 2 cats into 1 cat return L[aten.cat]( [ first, *rest, L[aten.slice](first, dim, 0, size), ], dim, ) else: # don't expect to hit this case, just fall back tmp = L[aten.cat](cat_input, dim) return L[aten.cat]( [ tmp, L[aten.slice](tmp, dim, 0, size), ], dim, ) def is_valid_splitwithsizes_cat(match): split_nodes = filter_nodes(match.nodes, aten.split_with_sizes) cat_nodes = filter_nodes(match.nodes, aten.cat) get_item_nodes = filter_nodes(match.nodes, operator.getitem) if len(split_nodes) != 1 or len(cat_nodes) != 1: return False split_node, cat_node = split_nodes[0], cat_nodes[0] # The dim of split and cat should match for passthrough if get_arg_value(split_node, 2, "dim") != get_arg_value(cat_node, 1, "dim"): return False get_item_args = { get_arg_value(get_item_node, 1) for get_item_node in get_item_nodes } assert None not in get_item_args split_sizes = get_arg_value(split_node, 1, "split_sizes") # All parts of split should be included in the cat if get_item_args != set(range(len(split_sizes))): return False # The order of get_item_args should same with cat_node used. # For example, if the split_node like split_with_sizes(input, [2, 2, 3], 1), # the cat node should be like cat([get_item(0), get_item(1), get_item(2)], 1). cat_items_args_order = [ get_arg_value(item_node, 1) for item_node in get_arg_value(cat_node, 0) ] if cat_items_args_order != list(range(len(split_sizes))): return False return True def same_meta(node1: torch.fx.Node, node2: torch.fx.Node): """True if two nodes have the same metadata""" val1 = node1.meta.get("val") val2 = node2.meta.get("val") return ( val1 is not None and val2 is not None and statically_known_true(sym_eq(val1.size(), val2.size())) and val1.layout == val2.layout and val1.dtype == val2.dtype and val1.device == val2.device and ( val1.layout != torch.strided or statically_known_true(sym_eq(val1.stride(), val2.stride())) ) ) noop_registry: Dict[Any, Any] = {} def register_noop_decomp(targets, nop_arg=0): def register_fun(cond): register_decomposition(targets, registry=noop_registry, unsafe=True)( (cond, nop_arg) # type: ignore[arg-type] ) return cond return register_fun @register_noop_decomp(aten.slice) def slice_noop(self, dim=0, start=None, end=None, step=1): if start is None or end is None: return False if ( statically_known_true(sym_eq(start, 0)) and statically_known_true(end >= 2**63 - 1) and statically_known_true(sym_eq(step, 1)) ): return True return False @register_noop_decomp(aten.slice_scatter, 1) def slice_scatter_noop(self, src, dim=0, start=None, end=None, step=1): if start is None: start = 0 if end is None: end = 2**63 - 1 if start == 0 and end >= 2**63 - 1 and step == 1: return True return False @register_noop_decomp(aten.repeat) def repeat_noop(self, repeats): return all(r == 1 for r in repeats) @register_noop_decomp(aten.constant_pad_nd) def constant_pad_nd(x, padding, fill_value=0): return all(p == 0 for p in padding) @register_noop_decomp(torch.ops.prims.convert_element_type) def convert_element_type_noop(x, dtype: torch.dtype): return x.dtype == dtype @register_noop_decomp(torch.ops.prims.device_put) def device_put_noop(x, device): return x.device == decode_device(device) @register_noop_decomp([aten.ceil, aten.floor, aten.round, aten.trunc]) def int_noop(x): return is_integer_dtype(x.dtype) @register_noop_decomp([aten.pow]) def pow_noop(a, b): return isinstance(b, int) and b == 1 @register_noop_decomp([aten.cat], lambda args: args[0][0]) def cat_noop(inputs, dim=0): return len(inputs) == 1 @register_noop_decomp(aten.view) def view_noop(arg, size): return arg.shape == size # Note, we also always have a check for identical metadata, which is why these # are safe @register_noop_decomp([aten.copy], nop_arg=1) @register_noop_decomp([aten.alias, aten.clone]) def true_noop(*args, **kwargs): return True def remove_noop_ops(graph: torch.fx.Graph): """ Removes both operations that are essentially aten.clone and operations that are essentially aten.alias from the graph. """ inputs = set() input_storages = set() output_storages = set() for node in graph.find_nodes(op="placeholder"): inputs.add(node) input_storages.add(get_node_storage(node)) output_node = next(iter(reversed(graph.nodes))) assert output_node.op == "output" outputs = output_node.args[0] if not isinstance(outputs, (list, tuple)): # nested subgraphs can have singleton outputs outputs = (outputs,) for out in outputs: if isinstance(out, torch.fx.Node): output_storages.add(get_node_storage(out)) for node in graph.nodes: if node.target in noop_registry: cond, src_index = noop_registry[node.target] if isinstance(src_index, int): src = node.args[src_index] else: src = src_index(node.args) if not isinstance(src, torch.fx.Node): continue # Don't introduce new aliasing between inputs and outputs. # See fx_passes/README.md for a discussion of why this is # necessary. node_storage = get_node_storage(node) src_storage = get_node_storage(src) node_is_view = node_storage == src_storage if ( not node_is_view and node_storage in output_storages and (src_storage in input_storages or src_storage in output_storages) ): continue # Even if input and outputs are expected to alias, # don't make "node is src" True if ( node_is_view and node in output_node.args and (src in inputs or src in output_node.args) ): continue is_valid, args, kwargs = get_fake_args_kwargs(node) if not is_valid: continue if same_meta(node, src) and cond(*args, **kwargs): node.replace_all_uses_with(src) graph.erase_node(node) def decompose_auto_functionalized(graph): """Decomposes auto_functionalized and triton_kernel_wrapper_functional nodes into clones and the underlying mutation node. We assume that the reinplacing pass runs before this; the reinplacing pass tells us (via rewriting the arguments or .meta to those nodes) which Tensors we should clone and which Tensors are safe to reinplace. """ graph_pass = PatternMatcherPass() @register_graph_pattern( CallFunctionVarArgs(torch.ops.higher_order.auto_functionalized), pass_dict=graph_pass, ) def _(match: Match, *args, **kwargs): from torch._higher_order_ops.auto_functionalize import auto_functionalized_dense only_clone_these_tensors = tuple( match.nodes[0].meta.get("only_clone_these_tensors", []) ) flat_args, spec = pytree.tree_flatten((args, kwargs)) # NB: we combine (args, kwargs) into flat args for replacing. # This is replace_by_example uses make_fx which does not support # tracing a function with kwargs. def decomp(*flat_args): args, kwargs = pytree.tree_unflatten(flat_args, spec) return auto_functionalized_dense(*args, only_clone_these_tensors, **kwargs) match.replace_by_example(decomp, flat_args, run_functional_passes=False) @register_graph_pattern( CallFunctionVarArgs(torch.ops.higher_order.triton_kernel_wrapper_functional), pass_dict=graph_pass, ) def _(match: Match, *args, **kwargs): from torch._higher_order_ops.triton_kernel_wrap import ( triton_kernel_wrapper_functional_dense, ) flat_args, spec = pytree.tree_flatten((args, kwargs)) # NB: we combine (args, kwargs) into flat args for replacing. # This is replace_by_example uses make_fx which does not support # tracing a function with kwargs. def decomp(*flat_args): args, kwargs = pytree.tree_unflatten(flat_args, spec) return (triton_kernel_wrapper_functional_dense(*args, **kwargs),) match.replace_by_example(decomp, flat_args, run_functional_passes=False) @register_graph_pattern( CallFunctionVarArgs(torch.ops.higher_order.auto_functionalized_v2), pass_dict=graph_pass, ) def _(match: Match, *args, **kwargs): from torch._higher_order_ops.auto_functionalize import ( auto_functionalized_v2_dense, ) only_clone_these_bases = tuple( match.nodes[0].meta.get("only_clone_these_tensors", []) ) flat_args, spec = pytree.tree_flatten((args, kwargs)) # NB: we combine (args, kwargs) into flat args for replacing. # This is replace_by_example uses make_fx which does not support # tracing a function with kwargs. def decomp(*flat_args): args, kwargs = pytree.tree_unflatten(flat_args, spec) return auto_functionalized_v2_dense(*args, only_clone_these_bases, **kwargs) match.replace_by_example(decomp, flat_args, run_functional_passes=False) graph_pass.apply(graph) for node in graph.find_nodes( op="call_function", target=torch.ops.higher_order.auto_functionalized ): raise AssertionError("auto_functionalized was not removed") for node in graph.find_nodes( op="call_function", target=torch.ops.higher_order.auto_functionalized_v2 ): raise AssertionError("auto_functionalized_v2 was not removed") for node in graph.find_nodes( op="call_function", target=torch.ops.higher_order.triton_kernel_wrapper_functional, ): raise AssertionError("triton_kernel_wrapper_functional was not removed") @register_lowering_pattern( CallFunction( aten.cat, ListOf( CallFunction( operator.getitem, CallFunction( aten.split_with_sizes, KeywordArg("input_"), Ignored(), Ignored(), _users=MULTIPLE, ), Ignored(), ), ), Ignored(), ), pass_number=2, extra_check=is_valid_splitwithsizes_cat, ) def splitwithsizes_cat_replace(match, input_): return input_ def is_valid_cat_splitwithsizes(match): cat_nodes = filter_nodes(match.nodes, aten.cat) split_nodes = filter_nodes(match.nodes, aten.split_with_sizes) if len(split_nodes) != 1 or len(cat_nodes) != 1: return False split_node, cat_node = split_nodes[0], cat_nodes[0] # the cat node has other users: can't eliminate if len(cat_node.users) > 1: return False # the dim of the cat and split should match dim = get_arg_value(split_node, 2, "dim") if dim != get_arg_value(cat_node, 1, "dim"): return False cat_inputs = list(get_arg_value(cat_node, 0)) split_sizes = get_arg_value(split_node, 1, "split_sizes") # the number of input tensors in cat and the # length of the split sizes should match if len(cat_inputs) != len(split_sizes): return False for cat_input, split_size in zip(cat_inputs, split_sizes): # each cat input tensor's size along dim # should match the corresponding split size if "val" not in cat_input.meta: return False cat_input_size = cat_input.meta["val"].size(dim) if cat_input_size != split_size: return False return True @register_lowering_pattern( CallFunction( aten.split_with_sizes, CallFunction( aten.cat, KeywordArg("input_"), Ignored(), _users=MULTIPLE, ), Ignored(), Ignored(), ), pass_number=2, extra_check=is_valid_cat_splitwithsizes, ) def cat_splitwithsizes_replace(match, input_): return input_ def view_to_reshape(gm): """ Replace view ops in the GraphModule to reshape ops. """ for nd in gm.graph.find_nodes( op="call_function", target=torch.ops.aten.view.default ): nd.target = torch.ops.aten.reshape.default def should_prefer_unfused_addmm(match): inp = match.kwargs["inp"] if not inp.meta["val"].is_cuda: return False output = match.output_node() return all(is_pointwise_use(use) for use in output.users) @register_graph_pattern( CallFunction(aten.addmm, KeywordArg("inp"), Arg(), Arg()), pass_dict=pass_patterns[2], extra_check=should_prefer_unfused_addmm, ) def unfuse_bias_add_to_pointwise(match: Match, mat1, mat2, *, inp): def repl(inp, x1, x2): return x1 @ x2 + inp match.replace_by_example(repl, [inp, mat1, mat2]) def is_valid_addmm_fusion(match): mat1, mat2 = match.args inp = match.kwargs["inp"] if not ( isinstance(inp, torch.fx.Node) and isinstance(inp.meta["val"], torch.Tensor) ): return False # Input is a number in_shape = inp.meta["val"].shape mm_shape = mat1.meta["val"].shape[0], mat2.meta["val"].shape[1] matched = is_expandable_to(in_shape, mm_shape) if not matched: return False # Shape mismatch return not should_prefer_unfused_addmm(match) @register_graph_pattern( CallFunction( aten.add, CallFunction(aten.mm, Arg(), Arg()), KeywordArg("inp"), ), pass_dict=pass_patterns[2], extra_check=is_valid_addmm_fusion, ) @register_graph_pattern( CallFunction( aten.add, KeywordArg("inp"), CallFunction(aten.mm, Arg(), Arg()), ), pass_dict=pass_patterns[2], extra_check=is_valid_addmm_fusion, ) def addmm(match, mat1, mat2, *, inp): def repl(inp, mat1, mat2): return aten.addmm(inp, mat1, mat2) match.replace_by_example(repl, [inp, mat1, mat2]) def check_shape_cuda_and_fused_int_mm_mul_enabled(match): return ( config.force_fuse_int_mm_with_mul and len(getattr(match.args[2].meta.get("val"), "shape", [])) == 2 and getattr(match.args[2].meta.get("val"), "is_cuda", False) ) @register_lowering_pattern( CallFunction( prims.convert_element_type.default, CallFunction( aten.mul, CallFunction( aten._int_mm, Arg(), Arg(), ), Arg(), ), Arg(), ), check_shape_cuda_and_fused_int_mm_mul_enabled, ) @register_lowering_pattern( CallFunction( aten.mul, CallFunction( aten._int_mm, Arg(), Arg(), ), Arg(), ), check_shape_cuda_and_fused_int_mm_mul_enabled, ) def fused_int_mm_mul(match: Match, mat1, mat2, mat3, out_dtype=None): return inductor.kernel.mm.tuned_fused_int_mm_mul(mat1, mat2, mat3, out_dtype) def is_index_put_and_requires_h2d_sync_for_gpu_value(node): from torch.fx.operator_schemas import normalize_function if node.target not in [ torch.ops.aten.index_put.default, torch.ops.aten.index_put_.default, ]: return False # Inductor falls back to aten.index_put_. # index_put_ will will call nonzero() and perform a H2D sync if # any of its indices are bool/byte tensors # However, it will short-circuit this H2D sync and run mask_fill_ # if the value we are putting is a cpu scalar. # Therefore, when inductor sees an index_put_ with byte tensor indices, # it should *not* convert the cpu scalar value into a gpu tensor. args_, kwargs_ = normalize_function(node.target, node.args, node.kwargs) # type: ignore[misc] any_byte_bool_indices = False indices = args_[1] for i in indices: if i is not None and i.meta["val"].dtype in [torch.bool, torch.int8]: any_byte_bool_indices = True val = args_[2].meta["val"] val_is_cpu_scalar = val.device.type == "cpu" and val.numel() == 1 # If both these conditions hold, then converting the val # to a gpu tensor will incur a H2D sync when inductor calls aten.index_put_ return any_byte_bool_indices and val_is_cpu_scalar class ConstructorMoverPass: def __init__(self, target: str, allow_outputs: bool = False) -> None: """ Move constructors from cpu to the target_device. Sweeps through the module, looking for constructor nodes that can be moved to the target_device. A constructor node can be moved to the target_device iff all of its users can also be moved (tested by cannot_be_moved). Otherwise, all dependent constructor nodes won't be moved. - target: target device type - allow_outputs: allow outputs to be moved """ self.target = target self.allow_outputs = allow_outputs assert isinstance(target, str), ( "target should be a string representing the device type. " f"Got: {type(target).__name__}" ) def allow_cpu_device(self, node: fx.Node) -> bool: """ Returns whether a node that returns a tensor on the target device may have cpu tensors as input. """ return node.target in ( torch.ops.aten.index.Tensor, torch.ops.aten.index_put.default, torch.ops.aten.index_put_.default, torch.ops.aten.copy.default, torch.ops.aten.copy_.default, torch.ops.aten.slice_scatter.default, ) def cannot_be_moved(self, node: fx.Node) -> bool: """ Returns whether a node can be moved to the target device. If this function returns False, it means that this node and all of its users won't be moved into the target device. """ if node.target == "output": return not self.allow_outputs if not ( isinstance(node.target, torch._ops.OpOverload) and node.target.namespace in ("prims", "aten") ): return True if is_index_put_and_requires_h2d_sync_for_gpu_value(node): return True return False def get_node_device(self, node: fx.Node) -> Optional[torch.device]: """ Get the device of a node. """ ten = node.meta.get("val") return None if not isinstance(ten, torch.Tensor) else ten.device def get_cpu_indeg_count(self, graph: fx.Graph) -> Dict[fx.Node, int]: """ Get the number of cpu inputs to a node """ cpu_indeg: Dict[fx.Node, int] = Counter() for node in graph.nodes: cpu_count = 0 def add_cpu_inp(node): nonlocal cpu_count device = self.get_node_device(node) cpu_count += device is not None and device.type == "cpu" pytree.tree_map_only(fx.Node, add_cpu_inp, (node.args, node.kwargs)) if cpu_count: cpu_indeg[node] = cpu_count return cpu_indeg def __call__(self, graph: fx.Graph) -> None: target_devices = set() constructors = [] for node in graph.nodes: device = self.get_node_device(node) if device and device.type == self.target: target_devices.add(device) if not ( isinstance(node.target, torch._ops.OpOverload) and node.target.namespace in ("prims", "aten") ): continue if not torch._subclasses.fake_tensor._is_tensor_constructor(node.target): continue if not node.kwargs.get("device") == torch.device("cpu"): continue constructors.append(node) # not handling multiple target devices initially if not constructors or len(target_devices) != 1: return movable_constructors = self.find_movable_constructors(graph, constructors) for node in movable_constructors: kwargs = node.kwargs.copy() kwargs["device"] = next(iter(target_devices)) node.kwargs = kwargs def find_movable_constructors( self, graph: fx.Graph, constructors: List[fx.Node] ) -> Set[fx.Node]: """ Starting from the cpu constructors, iterate through the graph and test that all of their downstream uses can safely be moved to cpu. """ cpu_indeg: Dict[fx.Node, int] = self.get_cpu_indeg_count(graph) # which constructors cannot be moved to gpu cannot_move_to_gpu: Set[fx.Node] = set() # For any node in the graph, which constructors does it have a dependency on constructor_dependencies: Dict[fx.Node, Set[fx.Node]] = defaultdict(set) # if a cpu node has a dependency on two different cpu constructors, # then if either constructor cannot be moved to gpu, the other cannot as well. # In this case any node with a dependency on one will have a dependency on the other equal_constructor_sets: Dict[fx.Node, Set[fx.Node]] = { c: {c} for c in constructors } def make_dependencies_equivalent( set1: Set[fx.Node], set2: Set[fx.Node] ) -> Set[fx.Node]: # could use union find but not worth complexity here set1.update(set2) for obj in set1: equal_constructor_sets[obj] = set1 return set1 queue: List[fx.Node] = list(constructors) for c in queue: constructor_dependencies[c].add(c) while queue: node = queue.pop() dependencies = constructor_dependencies[node] for user in node.users: if self.cannot_be_moved(user): cannot_move_to_gpu.update(dependencies) break # this node was used on a op which takes in multiple devices and output a gpu # tensor. we can convert its cpu input to gpu without making further changes node_device = self.get_node_device(user) if ( self.allow_cpu_device(user) and node_device and node_device.type == self.target ): del cpu_indeg[user] else: # otherwise, we should continue look at its downstream uses cpu_indeg[user] -= 1 if cpu_indeg[user] == 0: del cpu_indeg[user] queue.append(user) unioned_set = make_dependencies_equivalent( dependencies, constructor_dependencies[user] ) constructor_dependencies[user] = unioned_set for node in cpu_indeg: if constructor_dependencies[node]: cannot_move_to_gpu.update(constructor_dependencies[node]) all_cannot_move_to_gpu = cannot_move_to_gpu.copy() for constructor in cannot_move_to_gpu: all_cannot_move_to_gpu.update(equal_constructor_sets[constructor]) return set(constructors) - all_cannot_move_to_gpu def move_constructors_to_gpu(graph: fx.Graph) -> None: """ Moves intermediary tensors which are constructed on the cpu to gpu when safe """ ConstructorMoverPass(get_gpu_type())(graph)