For distribution: updated HDF5 1.10.0-alpha1

The HDF5 libraries and tools have been updated from a 1.9 series prototype to 1.10.0-alpha1. No other changes since v0.3.1.

The application has been rebuilt on RHEL6.7 and a tarball with the pre-built application is available to download.

Example to run the binary to write an outputfile of say 100GB:

mkdir test
cd test
tar -zxf swmr-testapp-0.3.1-RHEL6.7.tar.gz

# Run the writer with 4MB chunk sizes
./bin/swmr write -f data/img_4M.h5 -n 25600 --direct /path/to/output/file_4M.h5

# Run the writer with 16MB chunk sizes
./bin/swmr write -f data/img_16M.h5 -n 6400 --direct /path/to/output/file_16M.h5

For distribution: 4M and 16M chunk sizes

This is a minor patch to 0.3 where the example data files have been modified to provide both 4MB and 16MB datasets (previously there was only a 4MB dataset).

The application has been rebuilt on RHEL6.7 and a tarball with the pre-built application is available to download.

Example to run the binary to write an outputfile of say 100GB:

mkdir test
cd test
tar -zxf swmr-testapp-0.3.1-RHEL6.7.tar.gz

# Run the writer with 4MB chunk sizes
./bin/swmr write -f data/img_4M.h5 -n 25600 --direct /path/to/output/file_4M.h5

# Run the writer with 16MB chunk sizes
./bin/swmr write -f data/img_16M.h5 -n 6400 --direct /path/to/output/file_16M.h5

For distribution

This release is somewhat easier to distribute externally. The binary distribution is built on RHEL6.6 and require the following packages to be pre-installed on the system (both of which are available from RHEL repos): boost-program-options, apr-util, expat

Example to run the binary to write an outputfile of say 100GB:

mkdir test
cd test
tar -zxf swmr-testapp-0.3-RHEL6.6.tar.gz
./bin/swmr write -f data/full_chunk_img.h5 -n 25600 --direct /path/to/output/file.h5

Performance measurements

• The writer application has been improved to also provide some performance measurements, based on time stamping during the operation of the write loop.
• Chunking is now defined by the chunk settings of the test image.
• Added some basic optimisations of the HDF5 write configurations.

Initial basic test application

• Basic SWMR writer and reader functionality.
• The writer appends 2D images from a reference dataset to a 3D data volume.
• The reader monitors the growth of the dataset
  • For each update the reader will compare the latest frame against the reference frame.
  • A report of the comparisons is printed at the end.