Added HL examples
Description:
Examples existed for the HL lite, image, and table interfaces, but
were never included in the distro... until now!
Solution:
Added source files to CVS, updated Makefile and MANIFEST.
Platforms tested:
mir, heping, shanti
Improvement.
Description:
The t_cache takes a long long time to run and it tests HDF5 calls.
Move it to the back and let more basic tests to run first so that
basic features are tested first.
Platforms tested:
Tested in heping with pp.
Porting.
Description:
Red Storm's lustre used srand and rand to initialize their user mode
of lustre. HDF5 also uses srand and that messes up Lustre.
Solution:
Use -DRED_STORM to block out the HDF5 code that calls srand.
This is a temporary fix since Lustre has fixed its code from
calling srand and rand but the fix is not deployed at Red_storm
yet.
Platforms tested:
Tested at Red-Storm.
Maintenance on Windows
Description:
Remove project file blocktract.dsp according to Quincey's check in
Solution:
Platforms tested:
Windows
Misc. update:
Maintenance on Windows
Description:
Add in new project file fheapdll.dsp according to Quincey's check in
Solution:
Platforms tested:
Windows
Misc. update:
Maintenance on Windows
Description:
Update workspace according to Quincey's check in.
Remove blocktrack(dll).dsp, b_plus+tree(dll).dsp and sheap(dll).dsp and add in fheap(dll).dsp
Solution:
Platforms tested:
Windows XP
Misc. update:
new feature
Description
H5LTget_dataset_info did not allow for passing NULL in the arguments when that argument was not requested.:
Solution:
add conditions that allow to pass NULL in the return arguments, when one of them is not needed
Platforms tested:
linux
Misc. update:
Incrementtal checkin
Description:
Revise & update v2 B-tree code to separate the internal package-specific
routines from the library-callable "private" routines.
Similar updates for the fractal heap code.
Platforms tested:
FreeBSD 4.11 (sleipnir)
Mac OSX (amazon)
New feature
Description:
Check in initial "fractal heap" code, for supporting the group redesign.
Also, remove some remnants of the segmented heap/block tracker/B+tree code
which I didn't find during the last pass.
Platforms tested:
FreeBSD 4.11 (sleipnir)
Mac OSX (amazon)
Linux 2.4
Maintenance
Description:
Remove half-built (half-baked? :-) implementation of B+tree, block tracker
and segmented heaps. (In preparation for starting the new heap coding)
Platforms tested:
FreeBSD 4.11 (sleipnir)
Linux 2.4 (heping)
bug fix
Description:
ret_value is not set to SUCCEED in H5D_mpio_select_write
That causes the 64-bit intel compiler unhappy.
It will give a non-zero number and cause the testing fake failure.
Another one is picked up by cmpi again. For one optimization case,
another variable is not initialized properly and compiler set an unhappy number,
cause the test failed.
Solution:
Properly initialize those variables.
Platforms tested:
teragrid:parallel
mir:parallel
heping: parallel and sequential
tungsten:parallel
Misc. update:
h5committest doesn't finish due to no space left on device.
parallel tests still failed at tungsten with cmpi. It looks like it was a bug from cmpi.
bug fix for daily test
Description:
if (fm->select_chunk) is not NULL at sequential mode and not set by the library, it shouldn't be free.
Otherwise, it will cause failures in the daily test.
Solution:
Explicitly set (fm->select_chunk) to NULL for non-parallel case.
Platforms tested:
copper with parallel
heping with both sequential and parallel
shanti with sequential
Misc. update:
Hopefully only mir failed at daily test.
The previous check-in seems fixing the non-selection failure at heping.
Description:
release the special_collective_io_works macro at this platform.
Solution:
Platforms tested:
heping mpich 1.2.6
Misc. update:
Bug fix in collective IO code
Description:
When select none occurs in one process, fm->select_chunk hasn't been allocated the memory(not allocated until create_file_hyper); however, this needs to be checked; it was caught with cmpi at NCSA linux cluster.
Solution:
Relocate the place to allocate the memory space for fm->select_chunk.
Platforms tested:
cmpi at tungsten
mpich at heping
kind of minor and let daily test catch other errors.
Misc. update:
This won't fix the failure at mir.
bug fix
Description:
During the process of the combination of collective chunk IO code and the original IO code,
The return value of a function call hasn't been initialized. On 32-bit mode at modi4, that value isn't set to 0 by default. So a fake error value returns.
Solution:
Re-arrange the code so that the return value will be set properly.
Platforms tested:
parallel:
modi4 - 32bit
heping
seq:
heping
Misc. update:
Change to release script
Description:
Silenced doxygen's output when bin/release doc is run. Now if doxygen
has an error the release script will exit.
Someone forgot to run reconfigure after a Makefile.am, so testing the
release script seems to have had the side effect of regenerating a
Makefile.in.
Platforms tested:
mir
Change to release script
Description:
Silenced doxygen's output when bin/release doc is run. Now if doxygen
has an error the release script will exit.
Platforms tested:
mir
Purpose:
bug fix
Description:
modi4's sgi compiler is too old to support collective IO when there are no contributions from
some processes.
Solution:
Disable mpi_special_collective_io_work macro.
Platforms tested:
NCAR SGI
MODI4
COOPER
Misc. update:
code cleanup
Description:
remove two printf lines accidently added for debugging at NCSA cobalt.
Solution:
Platforms tested:
No need to test.
Misc. update:
Enhance collective chunk IO supports
Description:
Add a new test to check the correctness of the HDF5 library behavior for collective IO mode when one process doesn't have any contribution for IO.
Solution:
Platforms tested:
IBM AIX 5.2(copper)
Linux (heping) mpich-1.2.6
Misc. update:
Enhanced collective chunk IO supports
Description:
1. When using collective IO with chunking storage without any tuning, performance may become worse under some circumstances.
2. Current HDF5 handles raw-data IO per chunk. So for many small chunks, many small IOs will be passed into MPI-IO. That may cause bad performance.
3. For one IO per chunk case, sometimes performance with collective is worse than performance with independent. An obvious case is when only one process is doing IO and all other process are not doing IO, the collective IO will only add overheads for communication. We want to avoid this case. Some management inside our library needs to be done.
Solution:
- Added managements of collective IO supports for chunking storage
inside parallel HDF5
1) Implemented One IO with collective mode for all chunks in the
application by building one MPI derived datatype accross all
chunks.
2) Implemented the decision-making support to do collective IO inside
MPI-IO per chunk.
3) Added the decision-making support to do one IO accross all chunks
or to do multiple IOs with each IO per chunk.
4) Added the support to handle the case some processes won't do any IOs in
collectively.
5) Some MPI-IO package(mpich 1.2.6 or lower, e.g.) cannot handle
collective IO correctly for the case when some processes have no
contributions to IOs, a special macro is added to change
collective IO mode to independent IO mode inside HDF5 library.
Platforms tested:
Parallel:
IBM AIX 5.2(copper)
Linux (heping) mpich-1.2.6
SDSC Teragrid mpich-1.2.5
Linux(Tungsten) mpich-1.2.6
Altix(NCSA cobalt)
Seq:
Linux(heping)
Misc. update:
Batch file improvement
Description:
This batch file will be call by all of the tool test batch files
Solution:
Platforms tested:
Windows XP
Misc. update:
