[cig-commits] r14715 - doc/snac

Wed Apr 15 10:33:45 PDT 2009

Author: echoi
Date: 2009-04-15 10:33:45 -0700 (Wed, 15 Apr 2009)
New Revision: 14715

Modified:
   doc/snac/snac.lyx
Log:
1. Added a section, "Checkpointing and restarting".

2. Changed parameter description and the contents of post-processing according to the recent changes in checkpointing and restarting.



Modified: doc/snac/snac.lyx
===================================================================

--- doc/snac/snac.lyx	2009-04-15 16:25:17 UTC (rev 14714)
+++ doc/snac/snac.lyx	2009-04-15 17:33:45 UTC (rev 14715)
@@ -3469,6 +3469,38 @@
  
 \end_layout
 
+\begin_layout Description
+
+\family typewriter
+checkpointEvery
+\family default
+: the interval of time steps to checkpoint (see 
+\begin_inset CommandInset ref
+LatexCommand ref
+reference "sec:Checkpointing-and-Restarting"
+
+\end_inset
+
+).
+ 
+\end_layout
+
+\begin_layout Description
+
+\family typewriter
+restartTimestep
+\family default
+: the time step to restart (see 
+\begin_inset CommandInset ref
+LatexCommand ref
+reference "sec:Checkpointing-and-Restarting"
+
+\end_inset
+
+).
+ 
+\end_layout
+
 \begin_layout Subsection
 SNAC-Specific Parameters 
 \end_layout
@@ -3704,6 +3736,22 @@
  Used in tandem with SnacPlastic and SnacViscoPlastic, respectively.
 \end_layout
 
+\begin_layout Description
+
+\family typewriter
+SnacRestart
+\family default
+: required to restart a run.
+ See 
+\begin_inset CommandInset ref
+LatexCommand ref
+reference "sub:Restarting"
+
+\end_inset
+
+ for more details.
+\end_layout
+
 \begin_layout Standard
 Other plugins found in the plugins directory are experimental.
 \end_layout
@@ -5511,7 +5559,7 @@
 \begin_layout LyX-Code
 
 \size small
-<param name="maxTimeSteps"> 100001 </param> 
+<param name="maxTimeSteps"> 100000 </param> 
 \end_layout
 
 \begin_layout LyX-Code
@@ -6836,6 +6884,247 @@
 </StGermainData> 
 \end_layout
 
+\begin_layout Section
+Checkpointing and Restarting
+\begin_inset CommandInset label
+LatexCommand label
+name "sec:Checkpointing-and-Restarting"
+
+\end_inset
+
+
+\end_layout
+
+\begin_layout Subsection
+Chekpointing
+\end_layout
+
+\begin_layout Standard
+To restart an already finished run from a certain point in time, it is necessary
+ to record all the relevant information.
+ Generating such data set is called 
+\emph on
+checkpointing
+\emph default
+.
+ It is possible to checkpoint as a response to emergency like a certain
+ type of error or disk quota and wall clock limit soon to be exceeded.
+ However, SNAC currently only allows a user to specify the frequency of
+ checkpointnig by assigning a non-zero value to 
+\begin_inset Quotes sld
+\end_inset
+
+
+\family typewriter
+checkpointEvery
+\family default
+
+\begin_inset Quotes srd
+\end_inset
+
+ in an input file.
+ The meaning of the assigned value is the same with 
+\begin_inset Quotes sld
+\end_inset
+
+
+\family typewriter
+dumpEvery
+\family default
+
+\begin_inset Quotes srd
+\end_inset
+
+: 
+\emph on
+the number of time steps between events of checkpointing
+\emph default
+.
+ The default value of this parameter is 0 and SNAC checkpoints only at the
+ last time step.
+ Since the size of checkpoint data files are much larger than that of the
+ regular outputs, it is advised to set 
+\family typewriter
+checkpointEvery
+\family default
+ to be a sufficiently large value.
+\end_layout
+
+\begin_layout Standard
+Like dumping is logged in 
+\family typewriter
+timeStep.0
+\family default
+, checkpointing is logged in 
+\family typewriter
+checkpointTimeStep.0
+\family default
+.
+ This file is used when restart files are generated.
+\end_layout
+
+\begin_layout Standard
+Checkpoint data files have 
+\family typewriter
+CP
+\family default
+ attached to their names (e.g., 
+\family typewriter
+velocityCP.0
+\family default
+, 
+\family typewriter
+stressTensorCP.2
+\family default
+, etc).
+ The major difference from the regular outputs is the storing format of
+ element-associated variables such as stress and plastic strain.
+ To restart with the exactly the same state as before, those values need
+ to be recorded at the tetrahedra level.
+ On the contrary, the regular outputs contains one single-precision floating
+ point number per hexahedral element.
+\end_layout
+
+\begin_layout Subsection
+Restarting
+\begin_inset CommandInset label
+LatexCommand label
+name "sub:Restarting"
+
+\end_inset
+
+
+\end_layout
+
+\begin_layout Enumerate
+Run 
+\family typewriter
+snac2restart
+\family default
+ in the orignial 
+\family typewriter
+outputPath
+\family default
+ (see the next section for more details).
+ 
+\end_layout
+
+\begin_layout Enumerate
+Create a new directory for the restarting run.
+ This is mandatory to prevent accidental overwirting of the exisitng data
+ files.
+\end_layout
+
+\begin_layout Enumerate
+Move all the 
+\family typewriter
+*.restart
+\family default
+ files to the new directory.
+\end_layout
+
+\begin_layout Enumerate
+Prepare an input file that is identical with the one used for the original
+ run except that 
+\end_layout
+
+\begin_deeper
+\begin_layout Enumerate
+the new one has 
+\begin_inset Quotes sld
+\end_inset
+
+
+\family typewriter
+SnacRestart
+\family default
+
+\begin_inset Quotes srd
+\end_inset
+
+ in the extensions (=plugins) list 
+\emph on
+AND
+\end_layout
+
+\begin_layout Enumerate
+\begin_inset Quotes sld
+\end_inset
+
+
+\family typewriter
+restartTimestep
+\family default
+
+\begin_inset Quotes srd
+\end_inset
+
+ is set to be the time step to restart from 
+\emph on
+AND
+\end_layout
+
+\begin_layout Enumerate
+\begin_inset Quotes sld
+\end_inset
+
+
+\family typewriter
+outputPath
+\family default
+
+\begin_inset Quotes srd
+\end_inset
+
+ now points to where 
+\family typewriter
+*.restart
+\family default
+ files are.
+ SNAC will abort if it detects existing outputs there.
+\end_layout
+
+\end_deeper
+\begin_layout Enumerate
+Start running SNAC as usual but with the input file prepared as above.
+\end_layout
+
+\begin_layout Subsection
+
+\family typewriter
+snac2restart
+\end_layout
+
+\begin_layout Standard
+The general usage of 
+\family typewriter
+snac2restart
+\family default
+ is 
+\end_layout
+
+\begin_layout LyX-Code
+Usage: snac2restart [timeStep] [your "outputPath"] [path to write restart
+ files]
+\end_layout
+
+\begin_layout Standard
+timeStep is one of the checkpointed time steps and should be one of those
+ recorded in 
+\family typewriter
+checkpointTimeStep.0
+\family default
+.
+ The last two specifies a path to where your output files to process are
+ and where to write the processed restart files, respectively.
+ All the arguments are optional and by default 
+\family typewriter
+snac2restart
+\family default
+ tries to read and write in the current directory and process for the last
+ time step recorded inFiles with the suffix of *.restart are created.
+\end_layout
+
 \begin_layout Chapter
 Postprocessing and Graphics 
 \end_layout
@@ -6921,17 +7210,13 @@
 decompDims
 \family default
 .
- The resultant local element numbers in x, y, and z (lon, radius, lat) are
- written in the file 
+ The resultant global and local element numbers in x, y, and z as well as
+ decomposed processor numbers are written in the file 
 \family typewriter
-sim.{processor ID}
+sim.0
 \family default
-, which contents are common to all the processors.
- The data in 
-\family typewriter
-sim.*
-\family default
- are used to infer the number of processors in each decomposed axis.
+, which is common to all the processors.
+ 
 \end_layout
 
 \begin_layout Section
@@ -7163,26 +7448,17 @@
 \end_layout
 
 \begin_layout LyX-Code
-snac2vtk gnodex gnodey gnodez nprocx nprocy nprocz [time1 time2],
+snac2vtk path-to-outputs [time1 time2].
 \end_layout
 
 \begin_layout Standard
-where the meaning of arguments is the same as those for 
+The only required argument is a path to output fiels including 
 \family typewriter
-snac2dx
+sim.0
 \family default
- described in the previous section.
+, which contains critical information to process data.
  The last two optional arguments are used to set the range of time steps
  for data conversion.
- Note that the first three arguments are global 
-\emph on
-node
-\emph default
- numbers: they are always larger than corresponding global 
-\emph on
-element
-\emph default
- numbers by 1.
 \end_layout
 
 \begin_layout Standard
@@ -7199,30 +7475,32 @@
  to .vts files rather than contain actual data.
 \end_layout
 
+\begin_layout LyX-Code
+> snac2vtk ./ 
+\family roman
+(the last time step only)
+\end_layout
+
 \begin_layout Standard
-When the cookbook example was run on 2 
-\begin_inset Formula $\times$
-\end_inset
-
- 1 
-\begin_inset Formula $\times$
-\end_inset
-
- 2 processors, the right usage of 
-\family typewriter
-snac2vtk
-\family default
- is
+or
 \end_layout
 
 \begin_layout LyX-Code
-> snac2vtk 21 6 41 2 1 2
+> snac2vtk ./ 10 
+\family roman
+(from 10th to the last time step)
 \end_layout
 
 \begin_layout Standard
-.
+or
 \end_layout
 
+\begin_layout LyX-Code
+> snac2vtk ./ 1 1001 
+\family roman
+(from 1 to 1001th time step)
+\end_layout
+
 \begin_layout Paragraph*
 Example list of SNAC output files in the VTK format: 8 processors and 51
  outputs per processor.

