Actions

NCCHUNKIE: Difference between revisions

From BAWiki

(NCMERGE added as post-processor)
(NCMERGE added as pre-processor)
Line 35: Line 35:
Some concepts published in [https://support.hdfgroup.org/pubs/papers/2008-06_netcdf4_perf_report.pdf https://support.hdfgroup.org/pubs/papers/2008-06_netcdf4_perf_report.pdf] were used.
Some concepts published in [https://support.hdfgroup.org/pubs/papers/2008-06_netcdf4_perf_report.pdf https://support.hdfgroup.org/pubs/papers/2008-06_netcdf4_perf_report.pdf] were used.


|preprocessor=[[DATACONVERT]], [[GRIDCONVERT]], [[NCANALYSE]], [[NCAGGREGATE]], [[NCCUTOUT]], [[NCDELTA]], [[NCDVAR]], [[NetCDF Operators]], [[UNTRIM2007]], [[UNTRIM2]]
|preprocessor=[[DATACONVERT]], [[GRIDCONVERT]], [[NCANALYSE]], [[NCAGGREGATE]], [[NCCUTOUT]], [[NCDELTA]], [[NCDVAR]], [[NCMERGE]], [[NetCDF Operators]], [[UNTRIM2007]], [[UNTRIM2]]
|postprocessor=[[NCANALYSE]], [[NCAGGREGATE]], [[NCDELTA]], [[NCDVAR]], [[NCMERGE]], [[NCCHUNKIE]], [[NCCUTOUT]], [[NC2TABLE]], [[NetCDF Operators]]
|postprocessor=[[NCANALYSE]], [[NCAGGREGATE]], [[NCDELTA]], [[NCDVAR]], [[NCMERGE]], [[NCCHUNKIE]], [[NCCUTOUT]], [[NC2TABLE]], [[NetCDF Operators]]
|language=Fortran95
|language=Fortran95

Revision as of 12:19, 11 January 2022

Basic Information

Name of Program

NCCHUNKIE

Version-Date

August 2020

Description-Date

January 2022

Catchwords

data conversion
postprocessor
automatic adjustment of number of data READ to chunk size of input data
automatic computation of chunk sizes for result variables to support orthogonal data access
parrallelization (collective IO) using MPI

Acknowledgment: This project took advantage of netCDF software developed by UCAR/Unidata (www.unidata.ucar.edu/software/netcdf/).

Short Description of Functionality

Program NCCHUNKIE can be used to chunk data stored in cf-netcdf.nc files:

  1. Chunk sizes are computed automatically, and all dimensions are chunked to support orthogonal data access;
  2. Resulting chunks sizes lie somewhere between Disc Block Size and Chunk Buffer Size;
  3. Online compression is used during storage of data (low level of compression used, level 1);
  4. A netCDF-4 file is created (serial version creates NetCDF4 classic model format);
  5. Parameters cache size and cache nelems used within netCDF-4 API are automatically determined.

Just in case HDF error is detected during read of a data record, the program tries to reconstruct the wanted data set from adjacent (in time) records for the same variable. This type of repair works for time dependent variables only.

Input-Files

  1. No input steering data file required (parameters in command line or interactive specification);
  2. UGRID CF NetCDF data set (file type CF-NETCDF.NC).

Output-Files

  1. UGRID CF NetCDF data set (file type CF-NETCDF.NC);
  2. informative printer file of program execution (file type NCCHUNKIE.sdr) with informations related to program execution, required time for READ and WRITE of data, effective data transfer rates.
  3. trace of program execution (file type NCCHUNKIE.trc)

Methodology

Some concepts published in https://support.hdfgroup.org/pubs/papers/2008-06_netcdf4_perf_report.pdf were used.

Program(s) to run before this Program

DATACONVERT, GRIDCONVERT, NCANALYSE, NCAGGREGATE, NCCUTOUT, NCDELTA, NCDVAR, NCMERGE, NetCDF Operators, UNTRIM2007, UNTRIM2

Program(s) to run after this Program

NCANALYSE, NCAGGREGATE, NCDELTA, NCDVAR, NCMERGE, NCCHUNKIE, NCCUTOUT, NC2TABLE, NetCDF Operators

Additional Information

Language

Fortran95

Additional software

---

Original Version

G. Lang

Maintenance

G. Lang, S. Spohr

Documentation/Literature


back to Program Descriptions


Overview