Searched refs:tfprof (Results 1 – 25 of 69) sorted by relevance
123
44 %unignore tensorflow::tfprof;45 %unignore tensorflow::tfprof::PrintModelAnalysis;46 %unignore tensorflow::tfprof::NewProfiler;47 %unignore tensorflow::tfprof::ProfilerFromFile;48 %unignore tensorflow::tfprof::DeleteProfiler;49 %unignore tensorflow::tfprof::AddStep;50 %unignore tensorflow::tfprof::SerializeToString;51 %unignore tensorflow::tfprof::WriteProfile;52 %unignore tensorflow::tfprof::Profile;
31 namespace tfprof { namespace116 tensorflow::tfprof::Options* opts) { in ParseCmdLine()129 if (pieces[i] == string(tensorflow::tfprof::kOptions[0])) { in ParseCmdLine()135 } else if (pieces[i] == tensorflow::tfprof::kOptions[1]) { in ParseCmdLine()141 } else if (pieces[i] == tensorflow::tfprof::kOptions[2]) { in ParseCmdLine()147 } else if (pieces[i] == tensorflow::tfprof::kOptions[3]) { in ParseCmdLine()153 } else if (pieces[i] == tensorflow::tfprof::kOptions[4]) { in ParseCmdLine()159 } else if (pieces[i] == tensorflow::tfprof::kOptions[5]) { in ParseCmdLine()165 } else if (pieces[i] == tensorflow::tfprof::kOptions[6]) { in ParseCmdLine()172 } else if (pieces[i] == tensorflow::tfprof::kOptions[7]) { in ParseCmdLine()[all …]
29 namespace tfprof {39 tensorflow::tfprof::Options* opts);
20 namespace tfprof {
24 namespace tfprof {
38 namespace tfprof {
4 * [Start `tfprof`](#start-tfprof)17 tfprof command line tool uses the following input:45 tensorflow.tfprof.OpLogProto (optional). A proto used to provide extra operation60 ### Start `tfprof`62 #### Build `tfprof` argument72 #### Start `tfprof` Interactive Mode argument74 # The following commands will start tfprof interactive mode.121 #### Start `tfprof` Non-interactive Mode. argument124 # Runs tfprof in one-shot.135 tfprof>[all …]
12 For example `_trainable_variables` is created automatically by tfprof Python18 tfprof> scope -account_type_regexes VariableV2 -max_depth 4 -select params48 * It must have `RegisterStatistics('flops')` defined in TensorFlow. tfprof53 shape is only known during runtime. tfprof can fill in the missing shape with58 * If no RunMetadata provided, tfprof count float_ops of each graph node once,59 even if it is defined in tf.while_loop. This is because tfprof doesn't know60 how many times are run statically. If RunMetadata provided, tfprof calculate68 tfprof> scope -min_float_ops 1 -select float_ops -account_displayed_op_only82 tfprof> op -min_float_ops 1 -select float_ops -account_displayed_op_only -order_by float_ops
3 * [Times in TensorFlow and tfprof](#times-in-tensorflow-and-tfprof)10 ### Times in TensorFlow and tfprof23 tfprof reports 3 execution times:48 tfprof> code -show_name_regexes seq2seq_attention.* -max_depth 10 -select micros -order_by micros75 tfprof> code -start_name_regexes .*_add_seq2seq.* -show_name_regexes seq2seq_attention.* -max_depth…84 tfprof> code -max_depth 5 -select micros -order_by micros -start_name_regexes .*_add_seq2seq.* -mi…118 tfprof> op -select micros,occurrence -order_by micros134 tfprof> op -select micros,device -order_by micros153 tfprof options allow users to generate timeline in some advanced ways.172 tfprof> scope -max_depth 30 -select micros -min_micros 100000 -order_by micros
8 tfprof> graph -max_depth 10000000 -step 0 -account_type_regexes .* -output timeline:outfile=<filena…27 tfprof> op -select bytes -order_by bytes42 tfprof> scope -order_by bytes -select bytes -min_bytes 10000000062 tfprof> code -max_depth 10 -select bytes -order_by bytes -start_name_regexes .*seq2seq.* -min_byte…
3 tfprof analyzes profiles and generates advice for common issues.28 tfprof --graph_path=graph.pbtxt \32 tfprof> advise78 There is no magic behind advise mode. tfprof builds the profiles first, then
30 # param_stats can be tensorflow.tfprof.GraphNodeProto or31 # tensorflow.tfprof.MultiGraphNodeProto, depending on the view.109 tfprof allows you to profile statistics across multiple steps.
5 For all tfprof views, the profiles are processed with the following procedures20 (e.g. MatMul, Conv2D). `tfprof` also considers device as operation type.103 …of the type regexes specified. tfprof allow user to define extra operation types for graph nodes t…
27 from tensorflow.contrib.tfprof import model_analyzer28 from tensorflow.contrib.tfprof import tfprof_logger30 from tensorflow.contrib.tfprof.model_analyzer import ProfileContext
1 # tfprof: TensorFlow Profiler and Beyond3 <h1>Please use `tf.profiler.xxx` instead of `tf.contrib.tfprof.xxx`</h1>
8 name = "tfprof",
30 with tf.contrib.tfprof.ProfileContext('/tmp/train_dir') as pctx:36 tfprof> op -select micros,bytes,occurrence -order_by micros53 with tf.contrib.tfprof.ProfileContext('/tmp/train_dir',77 with tf.contrib.tfprof.ProfileContext('/tmp/train_dir',106 tfprof> code -max_depth 1000 -show_name_regexes .*model_analyzer.*py.* -select micros -account_type…125 tfprof> scope -account_type_regexes VariableV2 -max_depth 4 -select params141 tfprof> op -select micros,bytes,occurrence -order_by micros159 tfprof> advise219 tfprof> graph -step -1 -max_depth 100000 -output timeline:outfile=<filename>241 tfprof> code -select accelerator_micros -max_depth 100000 -output pprof:outfile=<filename> -trim_n…
3 package tensorflow.tfprof;6 // Only used to pass tfprof options from Python to C++.
3 package tensorflow.tfprof;30 // float_ops is filled by tfprof Python API when called. It requires the37 // Used to support tfprof "code" view.
18 namespace tfprof { namespace
23 namespace tfprof {
26 namespace tfprof {
70 tfprof.CodeDef origin_stack = 3;77 tfprof.OpLogProto graph_traceback = 5;