Difference between revisions of "Main Page/PIO"

From Nekcem
Jump to navigationJump to search
 
(57 intermediate revisions by 2 users not shown)
Line 1: Line 1:
This is the document page for parallel I/O library developed in summer 2010 by Jing Fu and Misun Min.
+
This is the document page for parallel I/O library developed for [http://www.mcs.anl.gov/~mmin/nekcem.html NekCEM].
  
 
== Background ==
 
== Background ==
Line 5: Line 5:
 
:Binary (used for production, compact size), or ASCII (used for debugging, human-readable)
 
:Binary (used for production, compact size), or ASCII (used for debugging, human-readable)
  
== Usage Introduction: ==
+
== Usage Introduction ==
  
Generally, several new parallel I/O approaches were developed, namely #3, 4, 5, 6, 7, 8.
+
Users can use #1,2,3,4,5,6,7,8 in .rea file for a specific example, by parameter 81. <br>
These numbers were used to indicate which I/O to use, in .rea file for a specific example, by parameter 103.  
 
  
param(103) = 1 and param(103) = 2 mean it's using Fortran I/O library to write binary and ASCII files.
 
These two approaches would produce one file per processor and we will refer them as ''old libraries''.
 
param(103) = 3 means it's using C-POSIX I/O libraries to write binary files. <br>
 
----
 
param(103) = 4 is N1 binary case (syncIO to 1 file).
 
 
param(103) = 5 is NM binary case (syncIO to M files).
 
  
param(103) = 6 is NM1 binary case (rbIO to 1 file with M writers).
+
Several '''advanced parallel I/O''' algorithms based on MPI-IO library were developed. <br>
 +
The analysis on the performance for those approaches are detailed in ''Parallel I/O
 +
performance for application-level checkpointing on the Blue Gene/P, by Jing Fu et. al.'' [http://www.mcs.anl.gov/~mmin]
  
param(103) = 8 is NMM binary case (rbIO to M files with M writers).
+
  proc=processors
 +
  '''param(81) = 4''': collective IO for N proc to 1 file ---> mpi-binary-N1-xxx.vtk
 +
  '''param(81) = 5''': collective IO for N proc to multiple M-files ---> mpi-binary-NM-xxx.vtk
 +
  '''param(81) = 6''': reduced-blocking IO for N proc to 1 file with M writers ---> mpi-binary-NM1-xxx.vtk
 +
  '''param(81) = 7''': reduced-blocking IO for N proc to 1 file with M writers ---> mpi-ascii-NM1-xxx.vtk
 +
  '''param(81) = 8''': reduced-blocking IO for N proc to multiple M-files with M writers ---> mpi-binary-NMM-xxx.vtk
  
param(103) = -6 is NM1 ASCII case (in tests it's refered to "7" instead of "-6").
+
'''Note''' that param(82) and param(83) need to be set correctly in *.rea file.
 +
  '''param(80)''' = ''exact'' total number of fields to be written
 +
  '''param(82)''' = number of output files
 +
  '''param(83)''' = frequency of restart output files, 0 (no restart output), #nn (iostep*nn)
 +
  '''param(84)''' = invoked with dump_number from the name of the restarting output file
  
Notice that param(82) and param(83) need to be set correctly in *.rea file.
 
  
param(82) = number of groups, param(83) = max number of fields we are going to write.
+
----
 
+
For '''traditional I/O''' approaches based on one file per processor, using ''old libraries'', one can use
== Implementation Details: ==
 
 
 
=== Topology of code ===
 
  
These I/O functions were initiated from cem_out function of cem_dg.F (and cem_dg2.F). <br>
+
  '''param(81) = 2''': use Fortran I/O library (ASCII, VTK format) ---> ascii-xxx.vtk
Implementation of parallel I/O routine were defined in vtkbin.c and rbIO_nekcem.c, where vtkcommon.c and vtkcommon.h serve as a place to hold common functions as well as global variables. <br>
+
  '''param(81) = 3''': use C-POSIX I/O libraries (binary, VTK format) ---> binary-xxx.vtk
  
 +
  '''param(81) = 1''': use nek5000's old output format (ASCII) -> xxx.fld
  
 +
== Visualization ==
  
== To-do List ==
+
The output files with param(81)=2,3,4,5,6,7,8 can be visualized with '''ParaView''' and '''VisIt'''. <br>
 +
The output with param(81)=1 can be visualized with nek5000 visualization tool, '''postx''' <br>

Latest revision as of 22:36, 16 December 2011

This is the document page for parallel I/O library developed for NekCEM.

Background

File Format
Binary (used for production, compact size), or ASCII (used for debugging, human-readable)

Usage Introduction

Users can use #1,2,3,4,5,6,7,8 in .rea file for a specific example, by parameter 81.


Several advanced parallel I/O algorithms based on MPI-IO library were developed.
The analysis on the performance for those approaches are detailed in Parallel I/O performance for application-level checkpointing on the Blue Gene/P, by Jing Fu et. al. [1]

 proc=processors
 param(81) = 4: collective IO for N proc to 1 file ---> mpi-binary-N1-xxx.vtk 
 param(81) = 5: collective IO for N proc to multiple M-files ---> mpi-binary-NM-xxx.vtk
 param(81) = 6: reduced-blocking IO for N proc to 1 file with M writers ---> mpi-binary-NM1-xxx.vtk
 param(81) = 7: reduced-blocking IO for N proc to 1 file with M writers ---> mpi-ascii-NM1-xxx.vtk
 param(81) = 8: reduced-blocking IO for N proc to multiple M-files with M writers ---> mpi-binary-NMM-xxx.vtk 

Note that param(82) and param(83) need to be set correctly in *.rea file.

 param(80) = exact total number of fields to be written
 param(82) = number of output files
 param(83) = frequency of restart output files, 0 (no restart output), #nn (iostep*nn) 
 param(84) = invoked with dump_number from the name of the restarting output file 



For traditional I/O approaches based on one file per processor, using old libraries, one can use

 param(81) = 2: use Fortran I/O library (ASCII, VTK format) ---> ascii-xxx.vtk
 param(81) = 3: use C-POSIX I/O libraries (binary, VTK format) ---> binary-xxx.vtk 
 param(81) = 1: use nek5000's old output format (ASCII) -> xxx.fld

Visualization

The output files with param(81)=2,3,4,5,6,7,8 can be visualized with ParaView and VisIt.
The output with param(81)=1 can be visualized with nek5000 visualization tool, postx