j#6ddlZddlZddlZddlmZddlmZmZmZddl Z ddl m Z ddl m Z ddl mZddlmZdd lmZdd lmZejeZ ed Zd Zd ZedZedZedZeddZ dZ!d4dZ"de j#j$de%fdZ&dZ' d5dZ(ede(ee!dZ)ede)d Z*ed!e*d"Z+ed#e+ee$Z,ed%e,Gd&d'e-Z.Gd(d)e-Z/ed*e j0j1fd+Z2ed*e j0j1fd,Z3ed*e j0j1fd-Z4ed*e j0j1fd.Z5ej6Gd/d0Z7d*e j0j1fd1Z8Gd2d3Z9dS)6N) import_module)AnyListOptional)#min_cut_rematerialization_partition)_guards)config) ts_compile) aot_autograd)register_debug_backendc J|rtd||jS)Nz&eager backend ignoring extra kwargs %s)logwarningforward)gmfake_tensor_inputskwargss QC:\Projects\Historical\venv\Lib\site-packages\torch/_dynamo/backends/debugging.pyeagerrs( F .fn,sM  & &D    % % % %  r) contextlibr)r$r%rs` @rrr&s=%$$$$$ Irc L|rtd|fd}|S)Nz/eager_noexcept backend ignoring extra kwargs %sc~ |S#t$r*}tjjd|d}~wwxYw)Nz7Unexpected exception when running generated GraphModule) Exceptiontorch_dynamoexcTorchDynamoException)argsers rinnerzeager_noexcept..inner?sV 2t9    -#88I  s <%7<)rr)rrrr0s` reager_noexceptr18sA O EvNNN Lrc |rtd|ddlm}fd}||d|}||S)Nz3pre_dispatch_eager backend ignoring extra kwargs %sr)make_fxcPtjj|Srr*fx Interpreterrun)r.rs r runnable_gmz'pre_dispatch_eager..runnable_gmQs!x##B''+T22rT) pre_dispatch)rr"torch.fx.experimental.proxy_tensorr3print_readable)rrrr3r9pre_dispatch_gms` rpre_dispatch_eagerr>Js S I6RRR::::::33333>ggk===?QRO""$$$ rc \|rtd|ddlmfd}|S)Nz,eager_debug backend ignoring extra kwargs %sr)SchemaCheckModec5tjj|cdddS#1swxYwYdSrr5)r.r@rs rr0zeager_debug..innerds _   7 78''++/6 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7s&?AA)rr#torch._subclasses.schema_check_moder@)rrrr0r@s` @r eager_debugrCZsY L BFKKKCCCCCC 777777 Lrts)namec@tj|Sr)r*jitscript)rrs r torchscriptrIks 9  B  rc fd}d|_|S)Nchtj|Sr)r*r6r7 boxed_run)r.fx_gs rr8zboxed_nop..runrs'x##D))33D999rT _boxed_call)rMexample_inputsr8s` r boxed_noprQqs*:::::CO Jrc$fd}d|_|S)Nctj5tj|cdddS#1swxYwYdSr)r* _subclassesCrossRefFakeModer6r7rL)r.rM ignore_op_fns rr8z$fake_crossref_boxed_nop..runzs   / / = = > >8''--77== > > > > > > > > > > > > > > > > > >s2A  A$'A$TrN)rMrPrVr8s` ` rfake_crossref_boxed_noprWys0>>>>>>CO Jropreturnc|jdvS)N)atenprimsprim) namespace)rXs rignore_builtinsr_s <4 44rctjjjstStjjjdkrt StjjjdksJt jt tS)Nall custom_ops)rV) r* _functorchr fake_tensor_crossrefrQrW functoolspartialr_rr get_nop_funcrhse   " 7X   5 > >&&&;|KKKK !8WWWWrc ^t|pt|pttd||fi|S)NT) fw_compiler bw_compiler partition_fnkeep_inference_input_mutations)r rQr)rrrjrkrs r aot_eagerrnsT <,9,98'+        ( ( !'  ( ((rrn)rE compiler_fnT)rjrmaot_eager_default_partitionerc |rtd|ddlm}ddi}|dx}r||t j|5tttdtj td  ||cdddS#1swxYwYdS) Nz;aot_eager_decomp_partition backend ignoring extra kwargs %sr)CompilerBisectorunlift_effect_tokensTaot_eager_decomp_partitioncDtdS)Nztorch._inductor.compile_fx)rselect_decomp_tablergrrz,aot_eager_decomp_partition..s"=,$$!!##rinductor)compiler)rjrkdecompositionsrl) rr!torch._inductor.compiler_bisectorrrget_config_changeupdatefunctorch_configpatchr rhrerfr)rrrrrconfig_patchesbisect_changess rrtrtsM   I6   CBBBBB,d3N);;$~. n---   / / " "  |$$$$#*3j       " " " " " " " " " " " " " " " " " " " "s(ACC Crtc tjsdn tj}tj|5t||fi|cdddS#1swxYwYdS)Nrb)rd)r~rdrrt)rrr config_vals r#aot_eager_decomp_partition_crossrefrs  4 3  2  Z @ @ @LL)".@KKFKKLLLLLLLLLLLLLLLLLLsA  AAr)rjaot_tsceZdZdS)ReluCompileErrorN__name__ __module__ __qualname__rgrrrrDrrceZdZdS)TestingOnlyCompileErrorNrrgrrrrrrrrc\|jjD]}|jtjkrt |Sr)graphnodestargetr*relurrrPnodes rrelu_compile_error_TESTING_ONLYrs4## ;%* $ $" " % Irc|jjD]/}|jtjkrtj|_d|_0||S)N)FReluRuntimeError)rrrr*r_assertr. recompilers rrelu_runtime_error_TESTING_ONLYrsG44 ;%* $ $-DK3DILLNNN Irc|jjD]<}|jtjkr%tj|_|jddf|_=||S)Nrr )rrrr*raddr.rrs r relu_accuracy_error_TESTING_ONLYrsR** ;%* $ $)DK1q)DILLNNN Irch|jjD]}|jdkrn|S|D]}|jst|S)N call_function)rrrXis_leafr)rrPrts r#non_leaf_compile_error_TESTING_ONLYrs^ 7o % % E & **y *) ) * IrceZdZUdZeejjed<e ed<e ed<ee ed<e ed<dZ e eejj ed<dZe eejed <dZe eed <d efd ZdS) ExplainOutputzu This is the output of :func:`torch._dynamo.explain()` There is no reason to create this class directly. graphs graph_countgraph_break_count break_reasonsop_countN ops_per_graph out_guards compile_timesrYc.d|jd}|d|jdz }|d|jdz }|dz }t|jD]7\}}|d|dzdz }|d |jdz }|d z }|jD] }|d |dz } 8|j9|d z }t|jD]\}}|d |dzdz }|D] }|d|dz } |j@|dz }t|jD]&\}}|d|dzdz }|dt|z }'|j |d|j dz }|S)Nz Graph Count:  zGraph Break Count: z Op Count: zBreak Reasons: z Break Reason r z: z Reason: z User Stack: z zOps per Graph: z Ops z z Out Guards: z Guard zCompile Times: ) rrr enumeraterreason user_stackrrstrr) selfoutputidx break_reason frame_summaryopsrXiguards r__str__zExplainOutput.__str__4s5!1555B(>BBBB0t}0000$$!*4+=!>!> 5 5 C 2A222 2F <\%8<<< (:>>> >F r)rrr__doc__rr*r6 GraphModule__annotations__intrrrNoderrGuardrrrrgrrrr"s % &&&& MMM37M8D/077704Jgm,-444#'M8C='''rrc||d|jjD}|t|z }|||jjr||j|||||fS)a This function is a utility which processes a torch.fx.GraphModule and accumulates information about its ops, graph breaks, and other details. It is intended to be used by the ExplainWithBackend class and `torch._dynamo.explain()` to provide details from Dynamo's graph capture. Parameters: gm (torch.fx.GraphModule): The GraphModule to be processed. graphs (list): A list that accumulates all the GraphModules processed. op_count (int): The total count of operations in all GraphModules processed so far. ops_per_graph (list): A list that accumulates the operations of each GraphModule. break_reasons (list): A list that accumulates the reasons for breaks in each GraphModule. Returns: tuple: A tuple containing the processed GraphModule, the updated lists of graphs, operations per graph, and break reasons, and the updated operation count. c2g|]}|jdk |jS)r)rXr).0rs r z)_explain_graph_detail..hs& P P P4TW5O5O4;5O5O5Or)appendrrlencompile_subgraph_reason graph_break)rrrrrrs r_explain_graph_detailrSs( MM" P P28> P P PC CH !-9R7888 vx ==rcFeZdZdZddZdejjfdZde fdZ dS) ExplainWithBackenda This class is intended to be used as a backend for `torch.compile`. It is composable with other backends. When used in this way, it accumulates information about graph breaks, ops, and other info and provides a string representation summarizing this information. Attributes: backend (str): The name of the backend to use for optimization. graphs (list): A list of the graphs captured by TorchDynamo. op_count (int): The total number of operations in all optimized graphs. break_reasons (list): A list of graph break reasons with stack traces. Example Usage: def fn(x): x = torch.sigmoid(x) return x torch._dynamo.reset() eb = ExplainWithBackend("inductor") optimized_fn = torch.compile(fn, backend=eb) result = optimized_fn(torch.randn(5)) print(eb.output()) rYNc\ddlm}|||_g|_d|_g|_dS)Nr )lookup_backendr)registryrbackendrrr)rrrs r__init__zExplainWithBackend.__init__sA,,,,,,%~g..   rrct||j|jg|j\}|_|_}|_|||Sr)rrrrr)rrrP_s r__call__zExplainWithBackend.__call__sM@U  T]B0BA A =DK4+=||B///rcxt|j}t|j||dz |j|j}|S)Nr )rrrrr)rrrs rrzExplainWithBackend.outputsA$+&&  K  !O   M    r)rYN) rrrrrr*r6rrrrrgrrrrqsm0    058/0000       rrr)NN): dataclassesrelogging importlibrtypingrrrr*functorch.compilerrtorch._functorchr r~torch._functorch.compilersr commonr rr register_backend getLoggerrrrrrr1r>rCrIrQrW_ops OpOverloadboolr_rhrnrprtrrr)rrr6rrrrr dataclassrrrrgrrrs######&&&&&&&&&& AAAAAA777777111111 @@@@@@g!!  @@@$"       t   5 -5$5555XXX ( ( ( (ky9999 , $!!! (6S""": %3M L L L .3 * - - -hF3333      y        i   (<(<)= EH,@     --------`> >>>><1111111111r