The Altair Community is migrating to a new platform to provide a better experience for you. In preparation for the migration, the Altair Community is on read-only mode from October 28 - November 6, 2024. Technical support via cases will continue to work as is. For any urgent requests from Students/Faculty members, please submit the form linked here
Optimal placement of Materialize Data & Free Memory
Maven
Hi guys,
I am trying to optimize my workfows in terms of calculation time. Here is one example where I dont know to put "cleanup"-Operators to sped up the calcuation. Comments on where you would put
Free Memory, Materialize Data or other Operators that should speed up the process.
Markus
I am trying to optimize my workfows in terms of calculation time. Here is one example where I dont know to put "cleanup"-Operators to sped up the calcuation. Comments on where you would put
Free Memory, Materialize Data or other Operators that should speed up the process.
Best,
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<process version="5.0">
<context>
<input/>
<output/>
<macros/>
</context>
<operator activated="true" class="process" compatibility="5.0.11" expanded="true" name="Root">
<parameter key="logverbosity" value="notes"/>
<process expanded="true" height="505" width="1217">
<operator activated="true" class="generate_data" compatibility="5.0.11" expanded="true" height="60" name="Generate Data" width="90" x="45" y="30">
<parameter key="number_examples" value="300"/>
<parameter key="number_of_attributes" value="10"/>
</operator>
<operator activated="true" class="discretize_by_user_specification" compatibility="5.0.11" expanded="true" height="94" name="Discretize" width="90" x="179" y="30">
<parameter key="attribute_filter_type" value="single"/>
<parameter key="attribute" value="label"/>
<parameter key="include_special_attributes" value="true"/>
<list key="classes">
<parameter key="ffirst" value="0.5"/>
<parameter key="last" value="Infinity"/>
</list>
</operator>
<operator activated="true" class="rename" compatibility="5.0.11" expanded="true" height="76" name="Rename" width="90" x="313" y="30">
<parameter key="old_name" value="att1"/>
<parameter key="new_name" value="id"/>
</operator>
<operator activated="true" class="set_role" compatibility="5.0.11" expanded="true" height="76" name="Set Role" width="90" x="447" y="30">
<parameter key="name" value="id"/>
<parameter key="target_role" value="id"/>
</operator>
<operator activated="true" class="loop_parameters" compatibility="5.0.0" expanded="true" height="94" name="ParameterIteration" width="90" x="581" y="30">
<list key="parameters">
<parameter key="Permutation.local_random_seed" value="[0.0;100000.0;999;linear]"/>
</list>
<process expanded="true" height="523" width="547">
<operator activated="true" class="subprocess" compatibility="5.0.0" expanded="true" height="76" name="Label Random Permutation (Y-Scrambling)" width="90" x="45" y="30">
<process expanded="true" height="521" width="1083">
<operator activated="true" class="multiply" compatibility="5.0.11" expanded="true" height="94" name="IOMultiplier (2)_1" width="90" x="45" y="30"/>
<operator activated="true" class="select_attributes" compatibility="5.0.0" expanded="true" height="76" name="AttributeFilter (3)" width="90" x="179" y="30">
<parameter key="attribute_filter_type" value="regular_expression"/>
<parameter key="regular_expression" value="label"/>
<parameter key="include_special_attributes" value="true"/>
</operator>
<operator activated="true" class="shuffle" compatibility="5.0.0" expanded="true" height="76" name="Permutation" width="90" x="313" y="30">
<parameter key="use_local_random_seed" value="true"/>
<parameter key="local_random_seed" value="22322"/>
</operator>
<operator activated="true" class="generate_id" compatibility="5.0.0" expanded="true" height="76" name="IdTagging (3)" width="90" x="447" y="30"/>
<operator activated="true" class="set_role" compatibility="5.0.0" expanded="true" height="76" name="ChangeAttributeRole (2)" width="90" x="179" y="165">
<parameter key="name" value="label"/>
</operator>
<operator activated="true" class="select_attributes" compatibility="5.0.0" expanded="true" height="76" name="AttributeFilter (4)" width="90" x="313" y="165">
<parameter key="attribute_filter_type" value="regular_expression"/>
<parameter key="regular_expression" value="label|id"/>
<parameter key="invert_selection" value="true"/>
<parameter key="include_special_attributes" value="true"/>
</operator>
<operator activated="true" class="generate_id" compatibility="5.0.0" expanded="true" height="76" name="IdTagging (4)" width="90" x="447" y="165"/>
<operator activated="true" class="join" compatibility="5.0.0" expanded="true" height="76" name="ExampleSetJoin (2)" width="90" x="581" y="30"/>
<operator activated="true" class="materialize_data" compatibility="5.0.11" expanded="true" height="76" name="Materialize Data" width="90" x="715" y="30"/>
<operator activated="true" class="free_memory" compatibility="5.0.11" expanded="true" height="76" name="Free Memory (2)" width="90" x="849" y="30"/>
<connect from_port="in 1" to_op="IOMultiplier (2)_1" to_port="input"/>
<connect from_op="IOMultiplier (2)_1" from_port="output 1" to_op="AttributeFilter (3)" to_port="example set input"/>
<connect from_op="IOMultiplier (2)_1" from_port="output 2" to_op="ChangeAttributeRole (2)" to_port="example set input"/>
<connect from_op="AttributeFilter (3)" from_port="example set output" to_op="Permutation" to_port="example set input"/>
<connect from_op="Permutation" from_port="example set output" to_op="IdTagging (3)" to_port="example set input"/>
<connect from_op="IdTagging (3)" from_port="example set output" to_op="ExampleSetJoin (2)" to_port="left"/>
<connect from_op="ChangeAttributeRole (2)" from_port="example set output" to_op="AttributeFilter (4)" to_port="example set input"/>
<connect from_op="AttributeFilter (4)" from_port="example set output" to_op="IdTagging (4)" to_port="example set input"/>
<connect from_op="IdTagging (4)" from_port="example set output" to_op="ExampleSetJoin (2)" to_port="right"/>
<connect from_op="ExampleSetJoin (2)" from_port="join" to_op="Materialize Data" to_port="example set input"/>
<connect from_op="Materialize Data" from_port="example set output" to_op="Free Memory (2)" to_port="through 1"/>
<connect from_op="Free Memory (2)" from_port="through 1" to_port="out 1"/>
<portSpacing port="source_in 1" spacing="0"/>
<portSpacing port="source_in 2" spacing="54"/>
<portSpacing port="sink_out 1" spacing="0"/>
<portSpacing port="sink_out 2" spacing="54"/>
</process>
</operator>
<operator activated="true" class="subprocess" compatibility="5.0.0" expanded="true" height="112" name="TrainingSet: Modell mit XValidierung" width="90" x="313" y="30">
<process expanded="true" height="523" width="897">
<operator activated="true" class="x_validation" compatibility="5.0.0" expanded="true" height="112" name="XValidation" width="90" x="45" y="30">
<parameter key="create_complete_model" value="true"/>
<parameter key="sampling_type" value="shuffled sampling"/>
<parameter key="local_random_seed" value="10"/>
<process expanded="true" height="523" width="165">
<operator activated="true" class="random_forest" compatibility="5.0.0" expanded="true" height="76" name="RandomForest" width="90" x="45" y="30">
<parameter key="number_of_trees" value="5"/>
<parameter key="criterion" value="information_gain"/>
<parameter key="minimal_leaf_size" value="1"/>
<parameter key="maximal_depth" value="10"/>
<parameter key="subset_ratio" value="-1.0"/>
<parameter key="local_random_seed" value="-1"/>
</operator>
<connect from_port="training" to_op="RandomForest" to_port="training set"/>
<connect from_op="RandomForest" from_port="model" to_port="model"/>
<connect from_op="RandomForest" from_port="exampleSet" to_port="through 1"/>
<portSpacing port="source_training" spacing="0"/>
<portSpacing port="sink_model" spacing="0"/>
<portSpacing port="sink_through 1" spacing="0"/>
<portSpacing port="sink_through 2" spacing="0"/>
</process>
<process expanded="true" height="523" width="300">
<operator activated="true" class="apply_model" compatibility="5.0.0" expanded="true" height="76" name="ModelApplier" width="90" x="45" y="30">
<list key="application_parameters"/>
<parameter key="create_view" value="true"/>
</operator>
<operator activated="true" class="performance_binominal_classification" compatibility="5.0.11" expanded="true" height="76" name="BinominalClassificationPerformance" width="90" x="180" y="30">
<parameter key="main_criterion" value="youden"/>
<parameter key="psep" value="true"/>
</operator>
<connect from_port="model" to_op="ModelApplier" to_port="model"/>
<connect from_port="test set" to_op="ModelApplier" to_port="unlabelled data"/>
<connect from_op="ModelApplier" from_port="labelled data" to_op="BinominalClassificationPerformance" to_port="labelled data"/>
<connect from_op="BinominalClassificationPerformance" from_port="performance" to_port="averagable 1"/>
<portSpacing port="source_model" spacing="0"/>
<portSpacing port="source_test set" spacing="0"/>
<portSpacing port="source_through 1" spacing="0"/>
<portSpacing port="source_through 2" spacing="0"/>
<portSpacing port="sink_averagable 1" spacing="0"/>
<portSpacing port="sink_averagable 2" spacing="0"/>
</process>
</operator>
<operator activated="true" class="log" compatibility="5.0.0" expanded="true" height="112" name="ProcessLog" width="90" x="179" y="30">
<list key="log">
<parameter key="youden" value="operator.XValidation.value.performance"/>
<parameter key="accuracy" value="operator.XValidation.value.performance1"/>
<parameter key="psep" value="operator.XValidation.value.performance3"/>
<parameter key="deviation" value="operator.XValidation.value.deviation"/>
</list>
<parameter key="sorting_type" value="top-k"/>
<parameter key="sorting_dimension" value="youden"/>
<parameter key="sorting_k" value="1000"/>
</operator>
<connect from_port="in 1" to_op="XValidation" to_port="training"/>
<connect from_op="XValidation" from_port="model" to_op="ProcessLog" to_port="through 1"/>
<connect from_op="XValidation" from_port="training" to_op="ProcessLog" to_port="through 2"/>
<connect from_op="XValidation" from_port="averagable 1" to_op="ProcessLog" to_port="through 3"/>
<connect from_op="ProcessLog" from_port="through 1" to_port="out 1"/>
<connect from_op="ProcessLog" from_port="through 2" to_port="out 2"/>
<connect from_op="ProcessLog" from_port="through 3" to_port="out 3"/>
<portSpacing port="source_in 1" spacing="0"/>
<portSpacing port="source_in 2" spacing="0"/>
<portSpacing port="sink_out 1" spacing="0"/>
<portSpacing port="sink_out 2" spacing="0"/>
<portSpacing port="sink_out 3" spacing="0"/>
<portSpacing port="sink_out 4" spacing="0"/>
</process>
</operator>
<connect from_port="input 1" to_op="Label Random Permutation (Y-Scrambling)" to_port="in 1"/>
<connect from_op="Label Random Permutation (Y-Scrambling)" from_port="out 1" to_op="TrainingSet: Modell mit XValidierung" to_port="in 1"/>
<connect from_op="TrainingSet: Modell mit XValidierung" from_port="out 1" to_port="result 1"/>
<connect from_op="TrainingSet: Modell mit XValidierung" from_port="out 2" to_port="result 2"/>
<connect from_op="TrainingSet: Modell mit XValidierung" from_port="out 3" to_port="performance"/>
<portSpacing port="source_input 1" spacing="0"/>
<portSpacing port="source_input 2" spacing="0"/>
<portSpacing port="sink_performance" spacing="0"/>
<portSpacing port="sink_result 1" spacing="36"/>
<portSpacing port="sink_result 2" spacing="0"/>
<portSpacing port="sink_result 3" spacing="0"/>
</process>
</operator>
<operator activated="true" class="log_to_data" compatibility="5.0.0" expanded="true" height="76" name="ProcessLog2ExampleSet" width="90" x="715" y="30"/>
<operator activated="true" class="store" compatibility="5.0.11" expanded="true" height="60" name="Store" width="90" x="849" y="30">
<parameter key="repository_entry" value="../Data/Y_Scrambling"/>
</operator>
<operator activated="true" class="free_memory" compatibility="5.0.0" expanded="true" height="76" name="MemoryCleanUp (2)" width="90" x="983" y="30"/>
<connect from_op="Generate Data" from_port="output" to_op="Discretize" to_port="example set input"/>
<connect from_op="Discretize" from_port="example set output" to_op="Rename" to_port="example set input"/>
<connect from_op="Rename" from_port="example set output" to_op="Set Role" to_port="example set input"/>
<connect from_op="Set Role" from_port="example set output" to_op="ParameterIteration" to_port="input 1"/>
<connect from_op="ProcessLog2ExampleSet" from_port="exampleSet" to_op="Store" to_port="input"/>
<connect from_op="Store" from_port="through" to_op="MemoryCleanUp (2)" to_port="through 1"/>
<connect from_op="MemoryCleanUp (2)" from_port="through 1" to_port="result 1"/>
<portSpacing port="source_input 1" spacing="0"/>
<portSpacing port="sink_result 1" spacing="0"/>
<portSpacing port="sink_result 2" spacing="144"/>
</process>
</operator>
</process>
Markus
0
Answers
I would appreciate if you could say something about where to put Materialize Data and free memory in general. Especially in combination with loops.
Best,
Markus
you can simply forget to use Free Memory anyway. It will only call the garbage collection, but java will call it anyway if it needs to. So forget about this.
The Materialize Data should be placed inside loops (except X-Validation) if you are going to create new attributes inside the loop that should be dropped right after the loop. Additionally you should use Materialize Data if you have created multiple views in preprocessing stacked over each other and the data will be read more than once or twice. Then materializing might speed up the calculation if it fits in memory...
Greetings,
Sebastian
But since you guys update the code, and make the code more efficient, this operator becomes less useful?
@ where to place
What I do is simply place them everywhere, and see if it helps.
Then remove everywhere where it does not help.
For example:
You have a process with a big loop.
Maybe a combination of parameter optimization, attribute selection, etc.
You might want to place Materialize Data & Free Memory operators after the apply model operator.
If you notice your memory usage goes up less fast, or stops going up at all, its good right?
I doubt you can give a good theoretical explanation on where to place these operators.
It is hard because of the interaction with the JVM, which is not fully deterministic.