U 9%ep@snddlZddlZddlZddlZddlmZddlmZddlm Z m Z m Z m Z m Z mZmZddlmZddlZddlmmZddlmZddlmZmZmZmZmZddlm Z m!Z!dd l"m#Z#m$Z$d d d d ddgZ%dZ&dd Z'GdddZ(Gdd d eZ)ddde*e*e*e*e*e ddd Z+e*e)dddZ,d e-ee-e.ddd Z/Gddde(Z0GdddZ1dS)!N)Enum)partial)AnyCallableDictIterableListOptionalTuple)warn)_get_privateuse1_backend_name)_add_execution_trace_observer!_disable_execution_trace_observer _enable_execution_trace_observer_ExperimentalConfig _remove_execution_trace_observer)kineto_availableProfilerActivity) MemoryProfileMemoryProfileTimelinesupported_activitiesProfilerActionscheduletensorboard_trace_handlerprofileExecutionTraceObserverZ ProfilerStepcCs tjS)a Returns a set of supported profiler tracing activities. Note: profiler uses CUPTI library to trace on-device CUDA kernels. In case when CUDA is enabled but CUPTI is not available, passing ``ProfilerActivity.CUDA`` to profiler results in using the legacy CUDA profiling code (same as in the legacy ``torch.autograd.profiler``). This, in turn, results in including CUDA time in the profiler table output, but not in the JSON trace. )torchautogradZ_supported_activitiesrrV/var/www/html/Darija-Ai-API/env/lib/python3.8/site-packages/torch/profiler/profiler.pyr#s c @seZdZdZddddddddeeeeeeeeeedddZ ddZ d d Z d d Z d dZ ddZedddZd+eedddZd,eedddZddZeeddd Zeedd!d"Zd#d$Zed%d&d'Zd-eeedd(d)d*ZdS)._KinetoProfileaLow-level profiler wrap the autograd profile Args: activities (iterable): list of activity groups (CPU, CUDA) to use in profiling, supported values: ``torch.profiler.ProfilerActivity.CPU``, ``torch.profiler.ProfilerActivity.CUDA``. Default value: ProfilerActivity.CPU and (when available) ProfilerActivity.CUDA. record_shapes (bool): save information about operator's input shapes. profile_memory (bool): track tensor memory allocation/deallocation. with_stack (bool): record source information (file and line number) for the ops. with_flops (bool): use formula to estimate the FLOPS of specific operators (matrix multiplication and 2D convolution). with_modules (bool): record module hierarchy (including function names) corresponding to the callstack of the op. e.g. If module A's forward call's module B's forward which contains an aten::add op, then aten::add's module hierarchy is A.B Note that this support exist, at the moment, only for TorchScript models and not eager mode models. experimental_config (_ExperimentalConfig) : A set of experimental options used by profiler libraries like Kineto. Note, backward compatibility is not guaranteed. .. note:: This API is experimental and subject to change in the future. Enabling shape and stack tracing results in additional overhead. When record_shapes=True is specified, profiler will temporarily hold references to the tensors; that may further prevent certain optimizations that depend on the reference count and introduce extra tensor copies. NF activities record_shapesprofile_memory with_stack with_flops with_modulesexperimental_configc Csb|r t|nt|_||_||_||_||_||_||_d|_ d|_ d|_ t }|dkr^||_ dS)NZ privateuseone) setrr"r#r&r$r%r'r(profilermem_tl use_devicer ) selfr"r#r$r%r&r'r(Zprivateuse1_backendrrr__init__Ps z_KinetoProfile.__init__cCs||dSN) prepare_trace start_tracer-rrrstartisz_KinetoProfile.startcCs |dSr/) stop_tracer2rrrstopmsz_KinetoProfile.stopc CsTtjtj|jktj|jktj|jkd|j|j|j |j |j d|j d |_ |j dS)NT) use_cudaZuse_cpuZuse_mtiar,r#r&r$r%r'Z use_kinetor()profrrCUDAr"ZCPUZMTIAr#r&r$r%r'r(r*Z_prepare_tracer2rrrr0ps    z_KinetoProfile.prepare_tracecCs|jdk st|j|jr*|dd|jr<|dd|jrN|dd|jr`|dd|jrr|ddt r| }|r|dt |t tdrd dlmm}|jjrdtjd <|d ddS) Nr$1r%r#r'r&ZdistributedInfo _inductorrZDISABLE_CUPTI_LAZY_REINIT)r*AssertionErrorZ _start_tracer$add_metadata_jsonr%r#r'r&r_get_distributed_infojsondumpshasattrrZtorch._inductor.configr:configZtritonZ cudagraphsosenviron)r-Z dist_infoZinductor_configrrrr1s*        z_KinetoProfile.start_tracecCs"|jdk st|jddddSr/)r*r;__exit__r2rrrr4sz_KinetoProfile.stop_trace)pathc Cs|js t|drtjdddd}||j|j}t|j(}t |d}| |W5QRXW5QRXt |j|S|j|SdS)zD Exports the collected trace in Chrome JSON format. .gzw+t.jsonFsuffixdeletewtN) r*r;endswithtempfileNamedTemporaryFilecloseexport_chrome_tracenameopengzip writelinesrBremove)r-rEfpZretvaluefinfoutrrrrQs    z"_KinetoProfile.export_chrome_traceself_cpu_time_total)rEmetriccCs|js t|j||S)aSave stack traces in a file in a format suitable for visualization. Args: path (str): save stacks file to this location; metric (str): metric to use: "self_cpu_time_total" or "self_cuda_time_total" .. note:: Example of using FlameGraph tool: - git clone https://github.com/brendangregg/FlameGraph - cd FlameGraph - ./flamegraph.pl --title "CPU time" --countname "us." profiler.stacks > perf_viz.svg )r*r; export_stacks)r-rEr[rrrr\s z_KinetoProfile.export_stacksr)group_by_input_shapegroup_by_stack_ncCs|js t|j||S)aAverages events, grouping them by operator name and (optionally) input shapes and stack. .. note:: To use shape/stack functionality make sure to set record_shapes/with_stack when creating profiler context manager. )r*r; key_averages)r-r]r^rrrr_s z_KinetoProfile.key_averagescCs|js t|jjS)z Returns the list of unaggregated profiler events, to be used in the trace callback or after the profiling is finished )r*r;Zfunction_eventsr2rrreventss z_KinetoProfile.events)keyvaluecCs&d|ddd}tj||dS)zo Adds a user defined metadata with a string key and a string value into the trace file "z\"N)replacerr_add_metadata_json)r-rarbZ wrapped_valuerrr add_metadatasz_KinetoProfile.add_metadatacCstj||dS)zs Adds a user defined metadata with a string key and a valid json value into the trace file N)rrre)r-rarbrrrr<sz _KinetoProfile.add_metadata_jsoncCs8ddlm}|r|s dS|||dS)Nr)backendZrankZ world_size)Ztorch.distributed distributed is_availableZis_initialized get_backendZget_rankZget_world_size)r-distrrrr=s z$_KinetoProfile._get_distributed_inforeturncsTd}fdd|D}|r.td|djdk rDjjdk sHttjjS)N)r#r$r%cs g|]}t|s|dqS)z=True)getattr).0ir2rr s z2_KinetoProfile._memory_profile..z, z required for memory profiling.) ValueErrorjoinr*Zkineto_resultsr;r)r-requiredmissingrr2r_memory_profiles z_KinetoProfile._memory_profile)rEdevicermc Cs |dkr"|jr"|jdkr"|jd}|dkr.schedule_fnrz#Invalid profiler schedule argumentsz>Profiler won't be using warmup, this can skew profiler results)rrr;r )rrrrrrrrrr/s )_rmcCstjS)zy Default profiler behavior - immediately starts recording the events, keeps doing it on every profiler step. )rr)rrrr_default_schedule_fnWsrF)dir_name worker_nameuse_gzipcs8ddlddlddlddfdd }|S)a  Outputs tracing files to directory of ``dir_name``, then that directory can be directly delivered to tensorboard as logdir. ``worker_name`` should be unique for each worker in distributed scenario, it will be set to '[hostname]_[pid]' by default. rNrlc sjsPzjddWn0tk rN}ztd|W5d}~XYnXsjddd}r|d}|j |dS)NT)exist_okzCan't create directory: r.z.pt.trace.jsonrF) rEisdirmakedirs Exception RuntimeError gethostnamegetpidtime_nsrQrs)r7e file_namerrBsockettimerrrr handler_fnls  z-tensorboard_trace_handler..handler_fn)rBrr)rrrrrrrr_s c seZdZdZddddddddddd eeeeeege feede fe e e e e ee ee d fddZ dd Zd d Zd d ZddZddZddZddZZS)raProfiler context manager. Args: activities (iterable): list of activity groups (CPU, CUDA) to use in profiling, supported values: ``torch.profiler.ProfilerActivity.CPU``, ``torch.profiler.ProfilerActivity.CUDA``. Default value: ProfilerActivity.CPU and (when available) ProfilerActivity.CUDA. schedule (Callable): callable that takes step (int) as a single parameter and returns ``ProfilerAction`` value that specifies the profiler action to perform at each step. on_trace_ready (Callable): callable that is called at each step when ``schedule`` returns ``ProfilerAction.RECORD_AND_SAVE`` during the profiling. record_shapes (bool): save information about operator's input shapes. profile_memory (bool): track tensor memory allocation/deallocation. with_stack (bool): record source information (file and line number) for the ops. with_flops (bool): use formula to estimate the FLOPs (floating point operations) of specific operators (matrix multiplication and 2D convolution). with_modules (bool): record module hierarchy (including function names) corresponding to the callstack of the op. e.g. If module A's forward call's module B's forward which contains an aten::add op, then aten::add's module hierarchy is A.B Note that this support exist, at the moment, only for TorchScript models and not eager mode models. experimental_config (_ExperimentalConfig) : A set of experimental options used for Kineto library features. Note, backward compatibility is not guaranteed. use_cuda (bool): .. deprecated:: 1.8.1 use ``activities`` instead. .. note:: Use :func:`~torch.profiler.schedule` to generate the callable schedule. Non-default schedules are useful when profiling long training jobs and allow the user to obtain multiple traces at the different iterations of the training process. The default schedule simply records all the events continuously for the duration of the context manager. .. note:: Use :func:`~torch.profiler.tensorboard_trace_handler` to generate result files for TensorBoard: ``on_trace_ready=torch.profiler.tensorboard_trace_handler(dir_name)`` After profiling, result files can be found in the specified directory. Use the command: ``tensorboard --logdir dir_name`` to see the results in TensorBoard. For more information, see `PyTorch Profiler TensorBoard Plugin `__ .. note:: Enabling shape and stack tracing results in additional overhead. When record_shapes=True is specified, profiler will temporarily hold references to the tensors; that may further prevent certain optimizations that depend on the reference count and introduce extra tensor copies. Examples: .. code-block:: python with torch.profiler.profile( activities=[ torch.profiler.ProfilerActivity.CPU, torch.profiler.ProfilerActivity.CUDA, ] ) as p: code_to_profile() print(p.key_averages().table( sort_by="self_cuda_time_total", row_limit=-1)) Using the profiler's ``schedule``, ``on_trace_ready`` and ``step`` functions: .. code-block:: python # Non-default profiler schedule allows user to turn profiler on and off # on different iterations of the training loop; # trace_handler is called every time a new trace becomes available def trace_handler(prof): print(prof.key_averages().table( sort_by="self_cuda_time_total", row_limit=-1)) # prof.export_chrome_trace("/tmp/test_trace_" + str(prof.step_num) + ".json") with torch.profiler.profile( activities=[ torch.profiler.ProfilerActivity.CPU, torch.profiler.ProfilerActivity.CUDA, ], # In this example with wait=1, warmup=1, active=2, repeat=1, # profiler will skip the first step/iteration, # start warming up on the second, record # the third and the forth iterations, # after which the trace will become available # and on_trace_ready (when set) is called; # the cycle repeats starting with the next step schedule=torch.profiler.schedule( wait=1, warmup=1, active=2, repeat=1), on_trace_ready=trace_handler # on_trace_ready=torch.profiler.tensorboard_trace_handler('./log') # used when outputting for tensorboard ) as p: for iter in range(N): code_iteration_to_profile(iter) # send a signal to the profiler that the next iteration has started p.step() NF) r"ron_trace_readyr#r$r%r&r'r(r6.c 's:|r t|nt} | dk rJtd| r4| tjntj| krJ| tjt| dks^tdt j ||||||| d|r||_ d|_ n t |_ d|_ ||_d|_| |j|_d|_tjtjfgtjtjf|jgtjtjf|j|jgtjtjf|j|jgtjtjfttd|j|jgtjtjfgtjtjf|jgtjtjf|jgtjtjfttd|jgtjtjfttd |jgtjtjfgtjtjfgtjtjf|j|jgtjtjf|j|j|jgtjtjf|j|j|j|jgtjtjf|j|j|j|jgtjdf|j|jgtjdf|j|jgtjdf|j|jgi|_tjt dS) Nz7use_cuda is deprecated, use activities argument insteadrz"No valid profiler activities foundr!TFz+Incorrect schedule: WARMUP followed by NONEz+Incorrect schedule: RECORD followed by NONEz-Incorrect schedule: RECORD followed by WARMUP)!r)rr addrr8rVlenr;superr.r record_stepsrrstep_numcurrent_action step_rec_fnrrrr0rr1rrr4 _trace_ready action_mapr7KinetoStepTrackerZinit_step_countPROFILER_STEP_NAME) r-r"rrr#r$r%r&r'r(r6Zactivities_set __class__rrr.s                     ?zprofile.__init__cCs ||Sr/)r3r2rrr __enter__\szprofile.__enter__cCs|tjtdSr/)r5r7rZerase_step_countr)r-exc_typeexc_valexc_tbrrrrD`szprofile.__exit__cCs:|tj|j|jr6tdt|j|_ |j dS)N ProfilerStep#) _transit_actionrrrrr7record_functionrrrrr2rrrr3ds  z profile.startcCs.|jr|jr|jddd||jddSr/)rrrDrrr2rrrr5ls z profile.stopcCs|jr|jr|jddd|j}|j}|jd7_||j|_|||jtj t |jrt dt ||_|j dS)zP Signals the profiler that the next profiling step has started. Nrr)rrrDrrrrr7rZincrement_steprrrr)r- prev_actionZcur_steprrrrqs  z profile.stepcCs|jr||dSr/)rr2rrrrszprofile._trace_readycCs(|j||f}|r$|D] }|qdSr/)rget)r-rrZ action_listactionrrrrszprofile._transit_action)r{r|r}r~r rrrrrrrrr.rrDr3r5rrr __classcell__rrrrr~s<q oc@sheZdZdZddZddZedddZd d Ze d d Z d dZ ddZ ddZ edddZdS)raExecution Trace Observer Each process can have a single ExecutionTraceObserver instance. The observer can be added to record function callbacks via calling register_callback() explicitly. Without calling unregister_callback(), repeated calls to register_callback() will not add additional observers to record function callbacks. Once an ExecutionTraceObserver is created, the start() and stop() methods control when the event data is recorded. Deleting or calling unregister_callback() will remove the observer from the record function callbacks, finalize the output file, and will stop incurring any overheads. cCsd|_d|_dS)z1 Initializes the default states. FN) _registered_execution_trace_runningr2rrrr.szExecutionTraceObserver.__init__cCs |dS)zO Calls unregister_callback() to make sure to finalize outputs. N)unregister_callbackr2rrr__del__szExecutionTraceObserver.__del__)output_file_pathcCs|js||_t||_dS)zz Adds ET observer to record function callbacks. The the data will be written to output_file_path. N)r_output_file_pathr )r-rrrrregister_callbacksz(ExecutionTraceObserver.register_callbackcCs|jr|td|_dS)zE Removes ET observer from record function callbacks. FN)rr5rr2rrrrsz*ExecutionTraceObserver.unregister_callbackcCs|jS)z^ Returns True if the execution trace observer is registered, otherwise False. )rr2rrr is_registeredsz$ExecutionTraceObserver.is_registeredcCs|jS)zK Returns True if the observer is running, otherwise False. )rr2rrr is_runningsz!ExecutionTraceObserver.is_runningcCs|jr|jstd|_dS)z$ Starts to capture. TN)rrrr2rrrr3s zExecutionTraceObserver.startcCs|jrtd|_dS)z# Stops to capture. FN)rrr2rrrr5szExecutionTraceObserver.stoprlcCs|jr |jStddS)z/ Returns the output file name. z^A callback to the ET profiler needs to be registered first before getting the output file pathN)rrrr2rrrget_output_file_paths z+ExecutionTraceObserver.get_output_file_pathN)r{r|r}r~r.rrrrpropertyrrr3r5rrrrrrs   )NF)2rTr>rBrNenumr functoolsrtypingrrrrrr r warningsr rZtorch.autograd.profilerrr*r7Ztorch._Cr Ztorch._C._profilerr rrrrZtorch.autogradrrZtorch.profiler._memory_profilerrr__all__rrr rrrrrrrrrrrrrs^  $  t  (