sg?*FddlZddlZddlZddlmZddlmZmZmZddl Z ddl m Z ddl m Z ddl mZddlmZdd lmZdd lmZej.eZ ed Zed Zed ZedZeddZdZeee dZ ede eedZ!ede!dZ"ede"eeZ#ede#Gdde$Z%Gdd e$Z&ed!e jNjPfd"Z)ed!e jNjPfd#Z*ed!e jNjPfd$Z+ed!e jNjPfd%Z,ejZGd&d'Z.d!e jNjPfd(Z/Gd)d*Z0y)+N) import_module)AnyListOptional)#min_cut_rematerialization_partition)_guards)config) ts_compile) aot_autograd)register_debug_backendc J|rtjd||jS)Nz&eager backend ignoring extra kwargs %s)logwarningforwardgmfake_tensor_inputskwargss S/var/www/html/venv/lib/python3.12/site-packages/torch/_dynamo/backends/debugging.pyeagerrs  .inner)sE t9  --##88I  s A*;A)rr)rrrr"s` reager_noexceptr#"s!  EvN Lrc |rtjd|ddlm}fd}||d|}|j |S)Nz3pre_dispatch_eager backend ignoring extra kwargs %sr)make_fxcZtjjj|SNrfx Interpreterrun)r rs r runnable_gmz'pre_dispatch_eager..runnable_gm;s#xx##B'++T22rT) pre_dispatch)rr"torch.fx.experimental.proxy_tensorr%print_readable)rrrr%r,pre_dispatch_gms` rpre_dispatch_eagerr14sE  I6R:3>gk=?QRO""$ rc R|rtjd|ddlmfd}|S)Nz,eager_debug backend ignoring extra kwargs %sr)SchemaCheckModec5tjjj|cdddS#1swYyxYwr'r()r r3rs rr"zeager_debug..innerNs;   788''+//6 7 7 7s +>A)rr#torch._subclasses.schema_check_moder3)rrrr"r3s` @r eager_debugr6Ds$  BFKC 7 Lrts)namec@tjj|Sr')rjitscript)rrs r torchscriptr<Us 99  B rc fd}d|_|S)Nc`tjjj|Sr')rr)r* boxed_run)r fx_gs rr+zboxed_nop..run\s#xx##D)33D99rT) _boxed_call)r@example_inputsr+s` r boxed_noprC[s:CO JrT) fw_compiler partition_fnkeep_inference_input_mutations aot_eager)r8 compiler_fn)rDrFaot_eager_default_partitionerc |rtjd|tjd5t t t dt jtd||cdddS#1swYyxYw)Nz;aot_eager_decomp_partition backend ignoring extra kwargs %sT)unlift_effect_tokensc4tdjS)Nztorch._inductor.compile_fx)rselect_decomp_tablerrz,aot_eager_decomp_partition..s=,$!!#rinductor)compiler)rD bw_compilerdecompositionsrE) rrfunctorch_configpatchr rC functoolspartialrrs raot_eager_decomp_partitionrXxst  I6    T : "  |!!$#**3j     " " " "s 8A11A:rX)rDaot_tsc eZdZy)ReluCompileErrorN__name__ __module__ __qualname__rNrrr[r[rr[c eZdZy)TestingOnlyCompileErrorNr\rNrrrbrbr`rrbrc|jjD]%}|jtjk(s!t |Sr')graphnodestargetrrelur[rrBnodes rrelu_compile_error_TESTING_ONLYrjs6# ;;%** $" "# Irc|jjD]<}|jtjk(s!tj |_d|_>|j|S)N)FReluRuntimeError)rdrerfrrg_assertr recompilerhs rrelu_runtime_error_TESTING_ONLYrosK4 ;;%** $--DK3DI4LLN Irc|jjD]K}|jtjk(s!tj |_|j ddf|_M|j|S)Nrr )rdrerfrrgaddr rnrhs r relu_accuracy_error_TESTING_ONLYrrsX* ;;%** $))DK1q)DI*LLN Irc|jjD]}|jdk(sn|S|D]}|jrt|S)N call_function)rdreopis_leafrb)rrBrits r#non_leaf_compile_error_TESTING_ONLYrxsT 77o %  *yy) )* IrceZdZUdZeej jed<e ed<e ed<ee ed<e ed<dZ e eej jed<dZe eej ed <dZe eed <d efd Zy) ExplainOutputzu This is the output of :func:`torch._dynamo.explain()` There is no reason to create this class directly. graphs graph_countgraph_break_count break_reasonsop_countN ops_per_graph out_guards compile_timesreturncd|jd}|d|jdz }|d|jdz }|dz }t|jD]C\}}|d|dzdz }|d |j dz }|d z }|j D] }|d |dz } E|j>|d z }t|jD]!\}}|d |dzdz }|D] }|d|dz } #|j?|dz }t|jD]"\}}|d|dzdz }|dt|z }$|j|d|jdz }|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: ) r|r}r enumerater~reason user_stackrrstrr) selfoutputidx break_reason frame_summaryopsruiguards r__str__zExplainOutput.__str__s !1!1 2"5'(>(>'?rBBJt}}oR00$$!*4+=+=!> 5 C Awc2 2F  \%8%8$9< >F r)r]r^r___doc__rrr) GraphModule__annotations__intrrrNoderrGuardrrrrNrrrzrzs %% && M37M8D/0704Jgmm,-4#'M8C='rrzcV|j||jjDcgc]}|jdk(s|j }}|t |z }|j||j jr|j|j |||||fScc}w)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. rt)appendrdrerurflencompile_subgraph_reason graph_break)rr{rrr~rirs r_explain_graph_detailrs( MM"#%88>> P4TWW5O4;; PC P CH !!--R778 vx == Qs B&B&cXeZdZdZddZdej jfdZde fdZ y) 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()) rNcTddlm}|||_g|_d|_g|_y)Nr )lookup_backendr)registryrbackendr{rr~)rrrs r__init__zExplainWithBackend.__init__8s',%g.   rrct||j|jg|j\}|_|_}|_|j ||Sr')rr{rr~r)rrrB_s r__call__zExplainWithBackend.__call__@sK@U  T]]B0B0BA =DK4+=||B//rct|j}t|j||dz |j|j}|S)Nr )rr{rzr~r)rr|rs rrzExplainWithBackend.outputFsB$++&  KK  !O    MM   r)rN) r]r^r_rrrr)rrrzrrNrrrrs.0 0588//0  rr)1 dataclassesrVlogging importlibrtypingrrrrfunctorch.compilerrtorch._functorchr rTtorch._functorch.compilersr commonr rr register_backend getLoggerr]rrr#r1r6r<rCrGrIrXrYrr[rbr)rrjrorrrx dataclassrzrrrNrrrs#&& A71 @g!  "     t   4#'  ky9 ,$! (6S"* %3M * -hF3  y  i (<(<(<(<)=)= EHH,@,@   ---`> ><11r