White Paper
3D
S.C
OM
© D
assa
ult S
ystè
me
s
Memory Management in CATIA V5
Generative Drafting Workbench
Version 1.0 – 3/1/2014
Written by Bruno Cailliere
Validated by: NA
Edited by: Anne Marie Miller
3D
S.C
OM
© D
assa
ult S
ystè
me
s
White Paper
2
3D
S.C
OM
© D
assa
ult S
ystè
me
s
Executive Summary
Generation of drawing views requires loading of all the related 3D documents.
Memory consumption increases due to loading of both 2D and 3D data. This
document will help users identify and manage such memory allocation issues
effectively through some best practices used in generative drafting workbench. It
gives information on generative drafting memory management according to the
scenario, the options, and the processed 3D data. The aim is to help users to choose
the best memory saving methodology according to their needs.
White Paper
3
3D
S.C
OM
© D
assa
ult S
ystè
me
s
Contents 1. Part 1. Diagnosis of Memory Problem in Generative Drafting......................................... 4
2. Part 2. General V5 Options to Reduce Memory Consumption ........................................ 6
2.1. Work with Cache System ........................................................................................... 6
2.2. Set the undo Stack Size to Zero ................................................................................. 6
2.3. No automatic backup option should be Activated: ...................................................... 7
3. Part 3. Best Practices..................................................................................................... 8
3.1. Choice of the View Generation mode ......................................................................... 9
3.2. Information of Memory Save in each View Generation Mode ................................... 10
3.3. Miscellaneous........................................................................................................... 12
4. Document History ........................................................................................................ 13
White Paper
4
3D
S.C
OM
© D
assa
ult S
ystè
me
s
1. Part 1. Diagnosis of Memory Problem in Generative
Drafting
While working on Generative Drafting workbench some of the following memory usage
problems are encountered by the users:
Warning messages are shown; e.g. “detected low virtual memory condition”, “detected a full
memory address space condition”.
Operating system error panel pop up indicating that there is no more memory available.
White Paper
5
3D
S.C
OM
© D
assa
ult S
ystè
me
s
CATIA crashes with or without any warning message. Abend trace (generated in the
CATTemp directory) file contains some information related to the memory usage, such as
“malloc failed...”. In the Stack Trace of Abend trace the memory percentage more than 70-
80% also indicates memory exhaustion problem. Refer to the image below.
In some cases, Memory problem may lead to a CPU loop. A CPU loop means that the
CATIA session is/seems to be frozen. CPU is used at 100% and action requested by the
CNEXT process may not be executed.
Memory Warning: Warning message appear as per the trigger set by user in tools- options.
This popup warns that because the amount of remaining memory is becoming low, user
should save your data and exit the session:
White Paper
6
3D
S.C
OM
© D
assa
ult S
ystè
me
s
2. Part 2. General V5 Options to Reduce Memory
Consumption
2.1. Work with Cache System Using a cache system considerably reduces the time required to load the data and
memory usage. In this mode, only a representation of the geometry is available and the
corresponding cgr file is loaded. This file does not contain actual design with all
specifications. Hence, this file is light weight and less memory consuming.
2.2. Set the undo Stack Size to Zero Stack size indicates the number of commands of global actions which can be undone for
each document. For undo to work, each operation is stored in the memory. Number of
these operations stored is defined by the stack size. If it is not set to 0, memory is used
to remember several operations. If the "undo" capability is not essential, set the “stack
size” to the minimum: 0.
White Paper
7
3D
S.C
OM
© D
assa
ult S
ystè
me
s
2.3. No automatic backup option should be Activated: If "warm start" capability is not essential, please activate the “No automatic backup”
option. With the "No automatic backup" option, documents are not saved automatically.
By Default, Automatic Back-up is activated. And with this option, a save operation
(frequency given in minutes) can be specified. Each save operation consumes memory.
Hence, activating "No automatic backup" option saves memory usage.
White Paper
8
3D
S.C
OM
© D
assa
ult S
ystè
me
s
3. Part 3. Best Practices In the following section, user can save memory by considering these points:
A. Choice of the view generation mode B. Information of memory save in each view generation mode C. Miscellaneous
There are four types for view generation modes:
1. Exact
2. Approximate
3. CGR
4. Raster
White Paper
9
3D
S.C
OM
© D
assa
ult S
ystè
me
s
3.1. Choice of the View Generation mode
Here is a simple algorithm which may help choose the type of view (check these items in the
given order):
Exact mode:
If generation of fillet edges or threads, or Exact 2D Geometry, or complex breakout, or
unfolded views is needed;
If associative dimensions are needed.
Raster mode:
If 2D geometry is not needed (if only an overview, without geometry dress-up is needed).
CGR mode:
If a precise 2D geometry computed from visualization data is needed.
Approximate mode:
If none of these item suits the need, the approximate mode may be the best to settle
memory allocation problems.
White Paper
10
3D
S.C
OM
© D
assa
ult S
ystè
me
s
3.2. Information of Memory Save in each View Generation Mode
Exact Mode
Exact mode is useful when the user requires section/breakout views, details Views,
Associative dimensions, Generation of 3D specifications (threads, fillets, dimensions, FT&A)
and Generation of Threads.
Memory optimizing tips related to various options in Exact mode.
1. De-activate the Exact preview for view generation. In this case, a part or product open in Visualization mode will not be loaded in Design mode for the preview, which optimizes memory consumption.
2. De-activate the Fillet representation modes and Generation of Axis-lines/Center-Lines/Threads if not needed. This prevents the generation of the above elements in the views. Hence memory required for view generation is reduced.
3. If hidden line representation is not asked, and many instances or bodies are hidden by
other, use the "Enable occlusion culling" option. This will load only the parts which will
be seen in the resulting view (instead of loading all of them, which is the case by default),
which optimizes memory consumption and CPU usage.
White Paper
11
3D
S.C
OM
© D
assa
ult S
ystè
me
s
4. Option "Only generate parts larger than" can be used to filter the smallest instances inside the whole product structure, if there is no need to load them.
5. Memory consumption according to various fillet options in Exact Mode
The Memory consumption is different for various types of fillets projection mode. Boundaries
projection requires the least memory and projected original Edges requires the maximum
memory in Fillet generation.
White Paper
12
3D
S.C
OM
© D
assa
ult S
ystè
me
s
Approximate
Approximate views are not as high in precision and quality as exact views.
This generation mode dramatically reduces memory consumption. Approximate mode has
advantage of section/breakout view creation. For restriction of approximate mode, please
refer to the documentation.
CGR
CGR views are not as high in quality as exact views, but they consume much less memory
during the generation. This may be useful when dealing with sophisticated products or
assemblies involving large amounts of data.
Raster Mode
Raster mode consumes very little memory. They do not load the parts in design mode. Use
Raster mode to get a 2D geometry image if only an overview is needed. Sections, dress-up
are not possible in raster view. Refer to the Documentation for more details.
3.3. Miscellaneous
1. For Part Drawings, use Part MML. It means copy the PartBody from the original
CATPart. Create a New CATPart and Paste Special (paste as result with link). Or
copy the .CATPart (or .model) document within a new CATProduct. In both above
cases, Geometry specifications of original Part are not needed for View Creation.
Hence Memory is saved.
2. Activate the Technological result for bodies containing threads. Thread generation in
views requires loading the whole .CATPart. Technological Results created in the
Part prevent loading of CATPart and thus save memory.
3. Since 3D data is not unloaded, memory usage increases after creating/updating
several views. Save, close the all documents, and re-open may help to release this
memory.
4. Unnecessary data may be loaded in addition by another workbench. Make sure that
these data are released before starting drafting workbench.
5. Starting a new session to execute the drafting scenario guarantees that the minimum
number of code libraries are loaded (no 3D application/workbench additional code).
6. Use the Batch Mode of update to run the update in the background and help user
carry on with other operations on the computer.
White Paper
13
3D
S.C
OM
© D
assa
ult S
ystè
me
s
4. Document History
Document Revision Date Revised By Changes/Notes
0.0 Jan 2009 A3R Original
1.0 Mar 2014 A3R Updated to new template
White Paper
14
3D
S.C
OM
© D
assa
ult S
ystè
me
s