odd h5diff behavior

Hdf-forum archives
1

I'm getting some output from h5diff which seems wrong.

[dws@oxygen02 history]$ /usr/local/contrib/hdf5-1.8.8-par/bin/h5diff -v -p 1e-30 ptclTrackTagFromAbsorber1s_HistoryOld.h5 /scr1_oxygen/dws/jenkins/workspace/vorpalall-build/qarnodes/oxygen/vptests/vpresults-oxygen/history/ptclTrackTagFromAbsorber1s/ptclTrackTagFromAbsorber1s_HistoryOld.h5 /trajectory
dataset: </trajectory> and </trajectory>
size: [339x7x6] [339x7x6]
position trajectory trajectory difference relative

···

------------------------------------------------------------------------
[ 0 1 0 ] 1.678294506e+243 0 1.678294506e+243 1
[ 0 1 1 ] 9.047791808e+164 0 9.047791808e+164 1
[ 0 1 2 ] 4.78541875e+175 0 4.78541875e+175 1
[ 0 1 3 ] 4.440349806e-38 0 4.440349806e-38 1
[ 0 1 4 ] 4.545801862e+97 0 4.545801862e+97 1
[ 0 1 5 ] 8.535925984e-307 0 8.535925984e-307 1
[ 0 2 0 ] 3.952525167e-322 0 3.952525167e-322 1
[ 0 2 1 ] 3.119382268e-319 0 3.119382268e-319 1
[ 0 2 2 ] 9.698666729e-317 0 9.698666729e-317 1
[ 0 2 3 ] 6.920039978e-310 0 6.920039978e-310 1
[ 0 2 4 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 2 5 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 3 0 ] 1.355073239e+248 0 1.355073239e+248 1
[ 0 3 2 ] 3.162020133e-322 0 3.162020133e-322 1
[ 0 3 3 ] 3.162020133e-322 0 3.162020133e-322 1
[ 0 3 5 ] 2.60988666e+180 0 2.60988666e+180 1
[ 0 4 0 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 4 1 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 4 2 ] 1.355073239e+248 0 1.355073239e+248 1
[ 0 4 5 ] 3.113058228e-319 0 3.113058228e-319 1
[ 0 5 0 ] 4.824065618e+228 0 4.824065618e+228 1
[ 0 5 1 ] 2.60988666e+180 0 2.60988666e+180 1
[ 0 5 2 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 5 3 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 5 4 ] 1.355073239e+248 0 1.355073239e+248 1

The diffs says that element (0,1,0) of the /trajectory dataset is 1.678294506e+243 (in bold above), but when I use h5dump to look at the dataset, I see 0s:

...
DATASET "trajectory" {
      DATATYPE H5T_IEEE_F64LE
      DATASPACE SIMPLE { ( 339, 7, 6 ) / ( H5S_UNLIMITED, H5S_UNLIMITED, H5S_UNLIMITED ) }
      DATA {
      (0,0,0): 1.09772e-05, 1.007e+12, 0, 0, 8, 0,
      (0,1,0): 0, 0, 0, 0, 0, 0,
      (0,2,0): 0, 0, 0, 0, 0, 0,
      (0,3,0): 0, 0, 0, 0, 0, 0,
      (0,4,0): 0, 0, 0, 0, 0, 0,
      (0,5,0): 0, 0, 0, 0, 0, 0,
      (0,6,0): 0, 0, 0, 0, 0, 0,
      (1,0,0): 1.10752e-05, 1.00993e+12, 0, 0, 8, 0,
      (1,1,0): 0, 0, 0, 0, 0, 0,
      (1,2,0): 0, 0, 0, 0, 0, 0,
      (1,3,0): 0, 0, 0, 0, 0, 0,
      (1,4,0): 0, 0, 0, 0, 0, 0,
      (1,5,0): 0, 0, 0, 0, 0, 0,
      (1,6,0): 0, 0, 0, 0, 0, 0,
...

Any ideas what's going on here. The datatype in both files is H5T_IEEE_F64LE, so I know this is not a datatype comparison issue.

Thanks,
Dave

2

Dave,

Could you please send your file to help@hdfgroup.org? We will take a look.

On which system are you running the command?

Thank you!

Elena

···

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Elena Pourmal The HDF Group http://hdfgroup.org
1800 So. Oak St., Suite 203, Champaign IL 61820
217.531.6112
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

On Jul 24, 2012, at 12:13 PM, Dave Wade-Stein wrote:

I'm getting some output from h5diff which seems wrong.

[dws@oxygen02 history]$ /usr/local/contrib/hdf5-1.8.8-par/bin/h5diff -v -p 1e-30 ptclTrackTagFromAbsorber1s_HistoryOld.h5 /scr1_oxygen/dws/jenkins/workspace/vorpalall-build/qarnodes/oxygen/vptests/vpresults-oxygen/history/ptclTrackTagFromAbsorber1s/ptclTrackTagFromAbsorber1s_HistoryOld.h5 /trajectory
dataset: </trajectory> and </trajectory>
size: [339x7x6] [339x7x6]
position trajectory trajectory difference relative
------------------------------------------------------------------------
[ 0 1 0 ] 1.678294506e+243 0 1.678294506e+243 1
[ 0 1 1 ] 9.047791808e+164 0 9.047791808e+164 1
[ 0 1 2 ] 4.78541875e+175 0 4.78541875e+175 1
[ 0 1 3 ] 4.440349806e-38 0 4.440349806e-38 1
[ 0 1 4 ] 4.545801862e+97 0 4.545801862e+97 1
[ 0 1 5 ] 8.535925984e-307 0 8.535925984e-307 1
[ 0 2 0 ] 3.952525167e-322 0 3.952525167e-322 1
[ 0 2 1 ] 3.119382268e-319 0 3.119382268e-319 1
[ 0 2 2 ] 9.698666729e-317 0 9.698666729e-317 1
[ 0 2 3 ] 6.920039978e-310 0 6.920039978e-310 1
[ 0 2 4 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 2 5 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 3 0 ] 1.355073239e+248 0 1.355073239e+248 1
[ 0 3 2 ] 3.162020133e-322 0 3.162020133e-322 1
[ 0 3 3 ] 3.162020133e-322 0 3.162020133e-322 1
[ 0 3 5 ] 2.60988666e+180 0 2.60988666e+180 1
[ 0 4 0 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 4 1 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 4 2 ] 1.355073239e+248 0 1.355073239e+248 1
[ 0 4 5 ] 3.113058228e-319 0 3.113058228e-319 1
[ 0 5 0 ] 4.824065618e+228 0 4.824065618e+228 1
[ 0 5 1 ] 2.60988666e+180 0 2.60988666e+180 1
[ 0 5 2 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 5 3 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 5 4 ] 1.355073239e+248 0 1.355073239e+248 1

The diffs says that element (0,1,0) of the /trajectory dataset is 1.678294506e+243 (in bold above), but when I use h5dump to look at the dataset, I see 0s:

...
DATASET "trajectory" {
      DATATYPE H5T_IEEE_F64LE
      DATASPACE SIMPLE { ( 339, 7, 6 ) / ( H5S_UNLIMITED, H5S_UNLIMITED, H5S_UNLIMITED ) }
      DATA {
      (0,0,0): 1.09772e-05, 1.007e+12, 0, 0, 8, 0,
      (0,1,0): 0, 0, 0, 0, 0, 0,
      (0,2,0): 0, 0, 0, 0, 0, 0,
      (0,3,0): 0, 0, 0, 0, 0, 0,
      (0,4,0): 0, 0, 0, 0, 0, 0,
      (0,5,0): 0, 0, 0, 0, 0, 0,
      (0,6,0): 0, 0, 0, 0, 0, 0,
      (1,0,0): 1.10752e-05, 1.00993e+12, 0, 0, 8, 0,
      (1,1,0): 0, 0, 0, 0, 0, 0,
      (1,2,0): 0, 0, 0, 0, 0, 0,
      (1,3,0): 0, 0, 0, 0, 0, 0,
      (1,4,0): 0, 0, 0, 0, 0, 0,
      (1,5,0): 0, 0, 0, 0, 0, 0,
      (1,6,0): 0, 0, 0, 0, 0, 0,
...

Any ideas what's going on here. The datatype in both files is H5T_IEEE_F64LE, so I know this is not a datatype comparison issue.

Thanks,
Dave

_______________________________________________
Hdf-forum is for HDF software users discussion.
Hdf-forum@hdfgroup.org
http://mail.hdfgroup.org/mailman/listinfo/hdf-forum_hdfgroup.org

3

Elena,

I've reduced it to a single hdf5 file, which, when h5diff'ed with itself, produces output similar to below.

I will send the file to the address below.

Thanks,
Dave

···

On Jul 24, 2012, at 11:25 AM, Elena Pourmal wrote:

Dave,

Could you please send your file to help@hdfgroup.org? We will take a look.

On which system are you running the command?

Thank you!

Elena
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Elena Pourmal The HDF Group http://hdfgroup.org
1800 So. Oak St., Suite 203, Champaign IL 61820
217.531.6112
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

On Jul 24, 2012, at 12:13 PM, Dave Wade-Stein wrote:

I'm getting some output from h5diff which seems wrong.

[dws@oxygen02 history]$ /usr/local/contrib/hdf5-1.8.8-par/bin/h5diff -v -p 1e-30 ptclTrackTagFromAbsorber1s_HistoryOld.h5 /scr1_oxygen/dws/jenkins/workspace/vorpalall-build/qarnodes/oxygen/vptests/vpresults-oxygen/history/ptclTrackTagFromAbsorber1s/ptclTrackTagFromAbsorber1s_HistoryOld.h5 /trajectory
dataset: </trajectory> and </trajectory>
size: [339x7x6] [339x7x6]
position trajectory trajectory difference relative
------------------------------------------------------------------------
[ 0 1 0 ] 1.678294506e+243 0 1.678294506e+243 1
[ 0 1 1 ] 9.047791808e+164 0 9.047791808e+164 1
[ 0 1 2 ] 4.78541875e+175 0 4.78541875e+175 1
[ 0 1 3 ] 4.440349806e-38 0 4.440349806e-38 1
[ 0 1 4 ] 4.545801862e+97 0 4.545801862e+97 1
[ 0 1 5 ] 8.535925984e-307 0 8.535925984e-307 1
[ 0 2 0 ] 3.952525167e-322 0 3.952525167e-322 1
[ 0 2 1 ] 3.119382268e-319 0 3.119382268e-319 1
[ 0 2 2 ] 9.698666729e-317 0 9.698666729e-317 1
[ 0 2 3 ] 6.920039978e-310 0 6.920039978e-310 1
[ 0 2 4 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 2 5 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 3 0 ] 1.355073239e+248 0 1.355073239e+248 1
[ 0 3 2 ] 3.162020133e-322 0 3.162020133e-322 1
[ 0 3 3 ] 3.162020133e-322 0 3.162020133e-322 1
[ 0 3 5 ] 2.60988666e+180 0 2.60988666e+180 1
[ 0 4 0 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 4 1 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 4 2 ] 1.355073239e+248 0 1.355073239e+248 1
[ 0 4 5 ] 3.113058228e-319 0 3.113058228e-319 1
[ 0 5 0 ] 4.824065618e+228 0 4.824065618e+228 1
[ 0 5 1 ] 2.60988666e+180 0 2.60988666e+180 1
[ 0 5 2 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 5 3 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 5 4 ] 1.355073239e+248 0 1.355073239e+248 1

The diffs says that element (0,1,0) of the /trajectory dataset is 1.678294506e+243 (in bold above), but when I use h5dump to look at the dataset, I see 0s:

...
DATASET "trajectory" {
      DATATYPE H5T_IEEE_F64LE
      DATASPACE SIMPLE { ( 339, 7, 6 ) / ( H5S_UNLIMITED, H5S_UNLIMITED, H5S_UNLIMITED ) }
      DATA {
      (0,0,0): 1.09772e-05, 1.007e+12, 0, 0, 8, 0,
      (0,1,0): 0, 0, 0, 0, 0, 0,
      (0,2,0): 0, 0, 0, 0, 0, 0,
      (0,3,0): 0, 0, 0, 0, 0, 0,
      (0,4,0): 0, 0, 0, 0, 0, 0,
      (0,5,0): 0, 0, 0, 0, 0, 0,
      (0,6,0): 0, 0, 0, 0, 0, 0,
      (1,0,0): 1.10752e-05, 1.00993e+12, 0, 0, 8, 0,
      (1,1,0): 0, 0, 0, 0, 0, 0,
      (1,2,0): 0, 0, 0, 0, 0, 0,
      (1,3,0): 0, 0, 0, 0, 0, 0,
      (1,4,0): 0, 0, 0, 0, 0, 0,
      (1,5,0): 0, 0, 0, 0, 0, 0,
      (1,6,0): 0, 0, 0, 0, 0, 0,
...

Any ideas what's going on here. The datatype in both files is H5T_IEEE_F64LE, so I know this is not a datatype comparison issue.

Thanks,
Dave

\

4

So it appears this problem with h5diff extends to h5dump as well.

I have an h5 file, which, when h5dump'ed on Win32 and Linux, produces different output.

Incorrect output is shown in bold.

Linux:
[dws@oxygen02 history]$ h5dump trajectory ptclTrackTagFromAbsorber2s_History.h5
...
   DATASET "trajectory" {
      DATATYPE H5T_IEEE_F64LE
      DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED, H5S_UNLIMITED ) }
      DATA {
      (0,0,0): 1.06204e-05, -4.79586e-07, 1.00833e+12, 1.87949e+08, 0, 8, 0,
      (0,1,0): 0, 0, 0, 0, 0, 0, 0,
      (0,2,0): 0, 0, 0, 0, 0, 0, 0,
      (0,3,0): 0, 0, 0, 0, 0, 0, 0,
      (0,4,0): 0, 0, 0, 0, 0, 0, 0,
      (0,5,0): 0, 0, 0, 0, 0, 0, 0,
      (0,6,0): 0, 0, 0, 0, 0, 0, 0,
      (0,7,0): 0, 0, 0, 0, 0, 0, 0,
      (0,8,0): 0, 0, 0, 0, 0, 0, 0,
      (0,9,0): 0, 0, 0, 0, 0, 0, 0,
      (0,10,0): 0, 0, 0, 0, 0, 0, 0,
      (0,11,0): 0, 0, 0, 0, 0, 0, 0,
      (1,0,0): 1.07179e-05, -4.79568e-07, 1.01124e+12, 1.89595e+08, 0, 8, 0,
      (1,1,0): 0, 0, 0, 0, 0, 0, 0,
      (1,2,0): 0, 0, 0, 0, 0, 0, 0,
      (1,3,0): 0, 0, 0, 0, 0, 0, 0,
      (1,4,0): 0, 0, 0, 0, 0, 0, 0,
      (1,5,0): 0, 0, 0, 0, 0, 0, 0,
      (1,6,0): 0, 0, 0, 0, 0, 0, 0,
      (1,7,0): 0, 0, 0, 0, 0, 0, 0,
      (1,8,0): 0, 0, 0, 0, 0, 0, 0,
      (1,9,0): 0, 0, 0, 0, 0, 0, 0,
      (1,10,0): 0, 0, 0, 0, 0, 0, 0,
      (1,11,0): 0, 0, 0, 0, 0, 0, 0,
...

Win32:
...
   DATASET "trajectory" {
      DATATYPE H5T_IEEE_F64LE
      DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED, H5S_UNLIMITED ) }
      DATA {
      (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0, 8,
      (0,0,6): 0,
      (0,1,0): 1.71412e+214, 1.73429e+185, 1.75754e+136, 3.44266e+218,
      (0,1,4): 5.73569e+169, 0, 1.34682e-308,
      (0,2,0): 1.91002e-308, 0, 0, 0, 0, 0, 0,
      (0,3,0): 0, 0, 0, 0, 0, 0, 0,
      (0,4,0): 0, 0, 0, 0, 0, 0, 0,
      (0,5,0): 0, 0, 0, 0, 0, 0, 0,
      (0,6,0): 0, 0, 0, 0, 0, 0, 0,
      (0,7,0): 0, 0, 0, 0, 0, 0, 0,
      (0,8,0): 0, 0, 0, 0, 0, 0, 0,
      (0,9,0): 0, 0, 0, 0, 0, 0, 0,
      (0,10,0): 0, 0, 0, 0, 0, 0, 0,
      (0,11,0): 0, 0, 0, 0, 0, 0, 0,
...

Oddly, h5dump produces still different output with the -d option. Here is the output of h5dump -d trajectory, run on Win32:

HDF5 "ptclTrackTagFromAbsorber2s_History.h5" {
DATASET "trajectory" {
   DATATYPE H5T_IEEE_F64LE
   DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED, H5S_UNLIMITED ) }
   DATA {
   (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0, 8, 0,
   (0,1,0): 0, 0, 0, 0, 0, 0, 0,
   (0,2,0): 0, 0, 0, 0, 0, 0, 0,
   (0,3,0): 0, 0, 0, 0, 0, 0, 0,
   (0,4,0): 0, 0, 0, 0, 0, 0, 0,
   (0,5,0): 0, 0, 0, 0, 0, 0, 0,
   (0,6,0): 0, 0, 0, 0, 0, 0, 0,
   (0,7,0): 0, 0, 0, 0, 0, 0, 0,
   (0,8,0): 0, 0, 0, 0, 0, 0, 0,
   (0,9,0): 0, 0, 0, 0, 0, 0, 0,
   (0,10,0): 0, 0, 0, 0, 0, 0, 0,
   (0,11,0): 0, 0, 0, 0, 0, 0, 0,
   (1,0,0): 1.07179e-005, -4.79568e-007, 1.01124e+012, 1.89595e+008, 0, 8, 0,
   (1,1,0): 0, 0, 0, 0, 0, 0, 0,
   (1,2,0): 0, 0, 0, 0, 0, 0, 0,
   (1,3,0): 1.37758e-308, 8.34708e-309, 4.17213e-308, 1.043e-308, 0, 0, 0,
   (1,4,0): 0, 0, 0, 0, 0, 0, 0,
   (1,5,0): 0, 0, 0, 0, 0, 0, 0,
   (1,6,0): 0, 0, 0, 0, 0, 0, 0,
   (1,7,0): 0, 0, 0, 0, 0, 0, 0,
   (1,8,0): 0, 0, 0, 0, 0, 0, 0,
   (1,9,0): 0, 0, 0, 0, 0, 0, 0,
   (1,10,0): 0, 0, 0, 0, 0, 0, 0,
   (1,11,0): 0, 0, 0, 0, 0, 0, 0,

I've tried versions 1.8.5 through 1.8.9, and they are all wrong, but in different ways!

Here is the output from 1.8.5-patch1, run on the same file on Win32:

   DATASET "trajectory" {
      DATATYPE H5T_IEEE_F64LE
      DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED, H5S_UNLIMITED ) }
      DATA {
      (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0, 8,
      (0,0,6): 0,
      (0,1,0): 0, 0, 0, 0, 0, 0, 0,
      (0,2,0): 0, 0, 1.32116e-308, 1.38263e-304, 0, 0, 0,
      (0,3,0): 0, 0, 0, 0, 0, 0, 0,
      (0,4,0): 0, 0, 0, 0, 0, 0, 0,
      (0,5,0): 0, 0, 0, 0, 0, 0, 0,
      (0,6,0): 0, 0, 0, 0, 0, 0, 0,
      (0,7,0): 0, 0, 0, 0, 0, 0, 0,
      (0,8,0): 0, 0, 0, 0, 0, 0, 0,
      (0,9,0): 0, 0, 0, 0, 0, 0, 0,
      (0,10,0): 0, 0, 1.32228e-308, 1.38263e-304, 0, 0, 0,
      (0,11,0): 0, 0, 0, 0, 0, 0, 0,

This is an anomalous case for us, as it only appears on this one file, out of 6200+ in our test suite. We can work around it for now, but it would be great to know what it is we are doing when we create this file that is causing the problem. It's small enough (144K) that I figured I'd just attached it so that anyone who wants to look at the file may do so.

Thanks!
Dave

ptclTrackTagFromAbsorber2s_History.h5 (141 KB)

···

I'm getting some output from h5diff which seems wrong.

[dws@oxygen02 history]$ /usr/local/contrib/hdf5-1.8.8-par/bin/h5diff -v -p 1e-30 ptclTrackTagFromAbsorber1s_HistoryOld.h5 /scr1_oxygen/dws/jenkins/workspace/vorpalall-build/qarnodes/oxygen/vptests/vpresults-oxygen/history/ptclTrackTagFromAbsorber1s/ptclTrackTagFromAbsorber1s_HistoryOld.h5 /trajectory
dataset: </trajectory> and </trajectory>
size: [339x7x6] [339x7x6]
position trajectory trajectory difference relative
------------------------------------------------------------------------
[ 0 1 0 ] 1.678294506e+243 0 1.678294506e+243 1
[ 0 1 1 ] 9.047791808e+164 0 9.047791808e+164 1
[ 0 1 2 ] 4.78541875e+175 0 4.78541875e+175 1
[ 0 1 3 ] 4.440349806e-38 0 4.440349806e-38 1
[ 0 1 4 ] 4.545801862e+97 0 4.545801862e+97 1
[ 0 1 5 ] 8.535925984e-307 0 8.535925984e-307 1
[ 0 2 0 ] 3.952525167e-322 0 3.952525167e-322 1
[ 0 2 1 ] 3.119382268e-319 0 3.119382268e-319 1
[ 0 2 2 ] 9.698666729e-317 0 9.698666729e-317 1
[ 0 2 3 ] 6.920039978e-310 0 6.920039978e-310 1
[ 0 2 4 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 2 5 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 3 0 ] 1.355073239e+248 0 1.355073239e+248 1
[ 0 3 2 ] 3.162020133e-322 0 3.162020133e-322 1
[ 0 3 3 ] 3.162020133e-322 0 3.162020133e-322 1
[ 0 3 5 ] 2.60988666e+180 0 2.60988666e+180 1
[ 0 4 0 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 4 1 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 4 2 ] 1.355073239e+248 0 1.355073239e+248 1
[ 0 4 5 ] 3.113058228e-319 0 3.113058228e-319 1
[ 0 5 0 ] 4.824065618e+228 0 4.824065618e+228 1
[ 0 5 1 ] 2.60988666e+180 0 2.60988666e+180 1
[ 0 5 2 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 5 3 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 5 4 ] 1.355073239e+248 0 1.355073239e+248 1

The diffs says that element (0,1,0) of the /trajectory dataset is 1.678294506e+243 (in bold above), but when I use h5dump to look at the dataset, I see 0s:

...
DATASET "trajectory" {
      DATATYPE H5T_IEEE_F64LE
      DATASPACE SIMPLE { ( 339, 7, 6 ) / ( H5S_UNLIMITED, H5S_UNLIMITED, H5S_UNLIMITED ) }
      DATA {
      (0,0,0): 1.09772e-05, 1.007e+12, 0, 0, 8, 0,
      (0,1,0): 0, 0, 0, 0, 0, 0,
      (0,2,0): 0, 0, 0, 0, 0, 0,
      (0,3,0): 0, 0, 0, 0, 0, 0,
      (0,4,0): 0, 0, 0, 0, 0, 0,
      (0,5,0): 0, 0, 0, 0, 0, 0,
      (0,6,0): 0, 0, 0, 0, 0, 0,
      (1,0,0): 1.10752e-05, 1.00993e+12, 0, 0, 8, 0,
      (1,1,0): 0, 0, 0, 0, 0, 0,
      (1,2,0): 0, 0, 0, 0, 0, 0,
      (1,3,0): 0, 0, 0, 0, 0, 0,
      (1,4,0): 0, 0, 0, 0, 0, 0,
      (1,5,0): 0, 0, 0, 0, 0, 0,
      (1,6,0): 0, 0, 0, 0, 0, 0,
...

Any ideas what's going on here. The datatype in both files is H5T_IEEE_F64LE, so I know this is not a datatype comparison issue.

Thanks,
Dave

5

Dave,

Thank you for your report!

I entered an issue into our database. We will try to resolve it by 1.8.11 release.

Elena

···

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Elena Pourmal The HDF Group http://hdfgroup.org
1800 So. Oak St., Suite 203, Champaign IL 61820
217.531.6112
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

On Oct 29, 2012, at 1:18 PM, Dave Wade-Stein wrote:

So it appears this problem with h5diff extends to h5dump as well.

I have an h5 file, which, when h5dump'ed on Win32 and Linux, produces different output.

Incorrect output is shown in bold.

Linux:
[dws@oxygen02 history]$ h5dump trajectory ptclTrackTagFromAbsorber2s_History.h5
...
   DATASET "trajectory" {
      DATATYPE H5T_IEEE_F64LE
      DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED, H5S_UNLIMITED ) }
      DATA {
      (0,0,0): 1.06204e-05, -4.79586e-07, 1.00833e+12, 1.87949e+08, 0, 8, 0,
      (0,1,0): 0, 0, 0, 0, 0, 0, 0,
      (0,2,0): 0, 0, 0, 0, 0, 0, 0,
      (0,3,0): 0, 0, 0, 0, 0, 0, 0,
      (0,4,0): 0, 0, 0, 0, 0, 0, 0,
      (0,5,0): 0, 0, 0, 0, 0, 0, 0,
      (0,6,0): 0, 0, 0, 0, 0, 0, 0,
      (0,7,0): 0, 0, 0, 0, 0, 0, 0,
      (0,8,0): 0, 0, 0, 0, 0, 0, 0,
      (0,9,0): 0, 0, 0, 0, 0, 0, 0,
      (0,10,0): 0, 0, 0, 0, 0, 0, 0,
      (0,11,0): 0, 0, 0, 0, 0, 0, 0,
      (1,0,0): 1.07179e-05, -4.79568e-07, 1.01124e+12, 1.89595e+08, 0, 8, 0,
      (1,1,0): 0, 0, 0, 0, 0, 0, 0,
      (1,2,0): 0, 0, 0, 0, 0, 0, 0,
      (1,3,0): 0, 0, 0, 0, 0, 0, 0,
      (1,4,0): 0, 0, 0, 0, 0, 0, 0,
      (1,5,0): 0, 0, 0, 0, 0, 0, 0,
      (1,6,0): 0, 0, 0, 0, 0, 0, 0,
      (1,7,0): 0, 0, 0, 0, 0, 0, 0,
      (1,8,0): 0, 0, 0, 0, 0, 0, 0,
      (1,9,0): 0, 0, 0, 0, 0, 0, 0,
      (1,10,0): 0, 0, 0, 0, 0, 0, 0,
      (1,11,0): 0, 0, 0, 0, 0, 0, 0,
...

Win32:
...
   DATASET "trajectory" {
      DATATYPE H5T_IEEE_F64LE
      DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED, H5S_UNLIMITED ) }
      DATA {
      (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0, 8,
      (0,0,6): 0,
      (0,1,0): 1.71412e+214, 1.73429e+185, 1.75754e+136, 3.44266e+218,
      (0,1,4): 5.73569e+169, 0, 1.34682e-308,
      (0,2,0): 1.91002e-308, 0, 0, 0, 0, 0, 0,
      (0,3,0): 0, 0, 0, 0, 0, 0, 0,
      (0,4,0): 0, 0, 0, 0, 0, 0, 0,
      (0,5,0): 0, 0, 0, 0, 0, 0, 0,
      (0,6,0): 0, 0, 0, 0, 0, 0, 0,
      (0,7,0): 0, 0, 0, 0, 0, 0, 0,
      (0,8,0): 0, 0, 0, 0, 0, 0, 0,
      (0,9,0): 0, 0, 0, 0, 0, 0, 0,
      (0,10,0): 0, 0, 0, 0, 0, 0, 0,
      (0,11,0): 0, 0, 0, 0, 0, 0, 0,
...

Oddly, h5dump produces still different output with the -d option. Here is the output of h5dump -d trajectory, run on Win32:

HDF5 "ptclTrackTagFromAbsorber2s_History.h5" {
DATASET "trajectory" {
   DATATYPE H5T_IEEE_F64LE
   DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED, H5S_UNLIMITED ) }
   DATA {
   (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0, 8, 0,
   (0,1,0): 0, 0, 0, 0, 0, 0, 0,
   (0,2,0): 0, 0, 0, 0, 0, 0, 0,
   (0,3,0): 0, 0, 0, 0, 0, 0, 0,
   (0,4,0): 0, 0, 0, 0, 0, 0, 0,
   (0,5,0): 0, 0, 0, 0, 0, 0, 0,
   (0,6,0): 0, 0, 0, 0, 0, 0, 0,
   (0,7,0): 0, 0, 0, 0, 0, 0, 0,
   (0,8,0): 0, 0, 0, 0, 0, 0, 0,
   (0,9,0): 0, 0, 0, 0, 0, 0, 0,
   (0,10,0): 0, 0, 0, 0, 0, 0, 0,
   (0,11,0): 0, 0, 0, 0, 0, 0, 0,
   (1,0,0): 1.07179e-005, -4.79568e-007, 1.01124e+012, 1.89595e+008, 0, 8, 0,
   (1,1,0): 0, 0, 0, 0, 0, 0, 0,
   (1,2,0): 0, 0, 0, 0, 0, 0, 0,
   (1,3,0): 1.37758e-308, 8.34708e-309, 4.17213e-308, 1.043e-308, 0, 0, 0,
   (1,4,0): 0, 0, 0, 0, 0, 0, 0,
   (1,5,0): 0, 0, 0, 0, 0, 0, 0,
   (1,6,0): 0, 0, 0, 0, 0, 0, 0,
   (1,7,0): 0, 0, 0, 0, 0, 0, 0,
   (1,8,0): 0, 0, 0, 0, 0, 0, 0,
   (1,9,0): 0, 0, 0, 0, 0, 0, 0,
   (1,10,0): 0, 0, 0, 0, 0, 0, 0,
   (1,11,0): 0, 0, 0, 0, 0, 0, 0,

I've tried versions 1.8.5 through 1.8.9, and they are all wrong, but in different ways!

Here is the output from 1.8.5-patch1, run on the same file on Win32:

   DATASET "trajectory" {
      DATATYPE H5T_IEEE_F64LE
      DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED, H5S_UNLIMITED ) }
      DATA {
      (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0, 8,
      (0,0,6): 0,
      (0,1,0): 0, 0, 0, 0, 0, 0, 0,
      (0,2,0): 0, 0, 1.32116e-308, 1.38263e-304, 0, 0, 0,
      (0,3,0): 0, 0, 0, 0, 0, 0, 0,
      (0,4,0): 0, 0, 0, 0, 0, 0, 0,
      (0,5,0): 0, 0, 0, 0, 0, 0, 0,
      (0,6,0): 0, 0, 0, 0, 0, 0, 0,
      (0,7,0): 0, 0, 0, 0, 0, 0, 0,
      (0,8,0): 0, 0, 0, 0, 0, 0, 0,
      (0,9,0): 0, 0, 0, 0, 0, 0, 0,
      (0,10,0): 0, 0, 1.32228e-308, 1.38263e-304, 0, 0, 0,
      (0,11,0): 0, 0, 0, 0, 0, 0, 0,

This is an anomalous case for us, as it only appears on this one file, out of 6200+ in our test suite. We can work around it for now, but it would be great to know what it is we are doing when we create this file that is causing the problem. It's small enough (144K) that I figured I'd just attached it so that anyone who wants to look at the file may do so.

Thanks!
Dave
<ptclTrackTagFromAbsorber2s_History.h5>

I'm getting some output from h5diff which seems wrong.

[dws@oxygen02 history]$ /usr/local/contrib/hdf5-1.8.8-par/bin/h5diff -v -p 1e-30 ptclTrackTagFromAbsorber1s_HistoryOld.h5 /scr1_oxygen/dws/jenkins/workspace/vorpalall-build/qarnodes/oxygen/vptests/vpresults-oxygen/history/ptclTrackTagFromAbsorber1s/ptclTrackTagFromAbsorber1s_HistoryOld.h5 /trajectory
dataset: </trajectory> and </trajectory>
size: [339x7x6] [339x7x6]
position trajectory trajectory difference relative
------------------------------------------------------------------------
[ 0 1 0 ] 1.678294506e+243 0 1.678294506e+243 1
[ 0 1 1 ] 9.047791808e+164 0 9.047791808e+164 1
[ 0 1 2 ] 4.78541875e+175 0 4.78541875e+175 1
[ 0 1 3 ] 4.440349806e-38 0 4.440349806e-38 1
[ 0 1 4 ] 4.545801862e+97 0 4.545801862e+97 1
[ 0 1 5 ] 8.535925984e-307 0 8.535925984e-307 1
[ 0 2 0 ] 3.952525167e-322 0 3.952525167e-322 1
[ 0 2 1 ] 3.119382268e-319 0 3.119382268e-319 1
[ 0 2 2 ] 9.698666729e-317 0 9.698666729e-317 1
[ 0 2 3 ] 6.920039978e-310 0 6.920039978e-310 1
[ 0 2 4 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 2 5 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 3 0 ] 1.355073239e+248 0 1.355073239e+248 1
[ 0 3 2 ] 3.162020133e-322 0 3.162020133e-322 1
[ 0 3 3 ] 3.162020133e-322 0 3.162020133e-322 1
[ 0 3 5 ] 2.60988666e+180 0 2.60988666e+180 1
[ 0 4 0 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 4 1 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 4 2 ] 1.355073239e+248 0 1.355073239e+248 1
[ 0 4 5 ] 3.113058228e-319 0 3.113058228e-319 1
[ 0 5 0 ] 4.824065618e+228 0 4.824065618e+228 1
[ 0 5 1 ] 2.60988666e+180 0 2.60988666e+180 1
[ 0 5 2 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 5 3 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 5 4 ] 1.355073239e+248 0 1.355073239e+248 1

The diffs says that element (0,1,0) of the /trajectory dataset is 1.678294506e+243 (in bold above), but when I use h5dump to look at the dataset, I see 0s:

...
DATASET "trajectory" {
      DATATYPE H5T_IEEE_F64LE
      DATASPACE SIMPLE { ( 339, 7, 6 ) / ( H5S_UNLIMITED, H5S_UNLIMITED, H5S_UNLIMITED ) }
      DATA {
      (0,0,0): 1.09772e-05, 1.007e+12, 0, 0, 8, 0,
      (0,1,0): 0, 0, 0, 0, 0, 0,
      (0,2,0): 0, 0, 0, 0, 0, 0,
      (0,3,0): 0, 0, 0, 0, 0, 0,
      (0,4,0): 0, 0, 0, 0, 0, 0,
      (0,5,0): 0, 0, 0, 0, 0, 0,
      (0,6,0): 0, 0, 0, 0, 0, 0,
      (1,0,0): 1.10752e-05, 1.00993e+12, 0, 0, 8, 0,
      (1,1,0): 0, 0, 0, 0, 0, 0,
      (1,2,0): 0, 0, 0, 0, 0, 0,
      (1,3,0): 0, 0, 0, 0, 0, 0,
      (1,4,0): 0, 0, 0, 0, 0, 0,
      (1,5,0): 0, 0, 0, 0, 0, 0,
      (1,6,0): 0, 0, 0, 0, 0, 0,
...

Any ideas what's going on here. The datatype in both files is H5T_IEEE_F64LE, so I know this is not a datatype comparison issue.

Thanks,
Dave

_______________________________________________
Hdf-forum is for HDF software users discussion.
Hdf-forum@hdfgroup.org
http://mail.hdfgroup.org/mailman/listinfo/hdf-forum_hdfgroup.org

6

The issue is uninitialized data in the dataset. Only the major
index(1,0,0-2,0,0-3,0,0-...) has data.

H5dump uses malloc to create the read buffer and the initial state of that
buffer is platform dependent. By changing this to a calloc call, the issue is
eliminated.

The question becomes how should h5dump behave?

Allen

···

On Monday, October 29, 2012 12:18:47 PM Dave Wade-Stein wrote:

So it appears this problem with h5diff extends to h5dump as well.

I have an h5 file, which, when h5dump'ed on Win32 and Linux, produces
different output.

Incorrect output is shown in bold.

Linux:
[dws@oxygen02 history]$ h5dump trajectory
ptclTrackTagFromAbsorber2s_History.h5 ...
   DATASET "trajectory" {
      DATATYPE H5T_IEEE_F64LE
      DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
H5S_UNLIMITED ) } DATA {
      (0,0,0): 1.06204e-05, -4.79586e-07, 1.00833e+12, 1.87949e+08, 0, 8, 0,
(0,1,0): 0, 0, 0, 0, 0, 0, 0,
      (0,2,0): 0, 0, 0, 0, 0, 0, 0,
      (0,3,0): 0, 0, 0, 0, 0, 0, 0,
      (0,4,0): 0, 0, 0, 0, 0, 0, 0,
      (0,5,0): 0, 0, 0, 0, 0, 0, 0,
      (0,6,0): 0, 0, 0, 0, 0, 0, 0,
      (0,7,0): 0, 0, 0, 0, 0, 0, 0,
      (0,8,0): 0, 0, 0, 0, 0, 0, 0,
      (0,9,0): 0, 0, 0, 0, 0, 0, 0,
      (0,10,0): 0, 0, 0, 0, 0, 0, 0,
      (0,11,0): 0, 0, 0, 0, 0, 0, 0,
      (1,0,0): 1.07179e-05, -4.79568e-07, 1.01124e+12, 1.89595e+08, 0, 8, 0,
(1,1,0): 0, 0, 0, 0, 0, 0, 0,
      (1,2,0): 0, 0, 0, 0, 0, 0, 0,
      (1,3,0): 0, 0, 0, 0, 0, 0, 0,
      (1,4,0): 0, 0, 0, 0, 0, 0, 0,
      (1,5,0): 0, 0, 0, 0, 0, 0, 0,
      (1,6,0): 0, 0, 0, 0, 0, 0, 0,
      (1,7,0): 0, 0, 0, 0, 0, 0, 0,
      (1,8,0): 0, 0, 0, 0, 0, 0, 0,
      (1,9,0): 0, 0, 0, 0, 0, 0, 0,
      (1,10,0): 0, 0, 0, 0, 0, 0, 0,
      (1,11,0): 0, 0, 0, 0, 0, 0, 0,
...

Win32:
...
   DATASET "trajectory" {
      DATATYPE H5T_IEEE_F64LE
      DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
H5S_UNLIMITED ) } DATA {
      (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0,
8, (0,0,6): 0,
      (0,1,0): 1.71412e+214, 1.73429e+185, 1.75754e+136, 3.44266e+218,
      (0,1,4): 5.73569e+169, 0, 1.34682e-308,
      (0,2,0): 1.91002e-308, 0, 0, 0, 0, 0, 0,
      (0,3,0): 0, 0, 0, 0, 0, 0, 0,
      (0,4,0): 0, 0, 0, 0, 0, 0, 0,
      (0,5,0): 0, 0, 0, 0, 0, 0, 0,
      (0,6,0): 0, 0, 0, 0, 0, 0, 0,
      (0,7,0): 0, 0, 0, 0, 0, 0, 0,
      (0,8,0): 0, 0, 0, 0, 0, 0, 0,
      (0,9,0): 0, 0, 0, 0, 0, 0, 0,
      (0,10,0): 0, 0, 0, 0, 0, 0, 0,
      (0,11,0): 0, 0, 0, 0, 0, 0, 0,
...

Oddly, h5dump produces still different output with the -d option. Here is
the output of h5dump -d trajectory, run on Win32:

HDF5 "ptclTrackTagFromAbsorber2s_History.h5" {
DATASET "trajectory" {
   DATATYPE H5T_IEEE_F64LE
   DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
H5S_UNLIMITED ) } DATA {
   (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0, 8,
0, (0,1,0): 0, 0, 0, 0, 0, 0, 0,
   (0,2,0): 0, 0, 0, 0, 0, 0, 0,
   (0,3,0): 0, 0, 0, 0, 0, 0, 0,
   (0,4,0): 0, 0, 0, 0, 0, 0, 0,
   (0,5,0): 0, 0, 0, 0, 0, 0, 0,
   (0,6,0): 0, 0, 0, 0, 0, 0, 0,
   (0,7,0): 0, 0, 0, 0, 0, 0, 0,
   (0,8,0): 0, 0, 0, 0, 0, 0, 0,
   (0,9,0): 0, 0, 0, 0, 0, 0, 0,
   (0,10,0): 0, 0, 0, 0, 0, 0, 0,
   (0,11,0): 0, 0, 0, 0, 0, 0, 0,
   (1,0,0): 1.07179e-005, -4.79568e-007, 1.01124e+012, 1.89595e+008, 0, 8,
0, (1,1,0): 0, 0, 0, 0, 0, 0, 0,
   (1,2,0): 0, 0, 0, 0, 0, 0, 0,
   (1,3,0): 1.37758e-308, 8.34708e-309, 4.17213e-308, 1.043e-308, 0, 0, 0,
   (1,4,0): 0, 0, 0, 0, 0, 0, 0,
   (1,5,0): 0, 0, 0, 0, 0, 0, 0,
   (1,6,0): 0, 0, 0, 0, 0, 0, 0,
   (1,7,0): 0, 0, 0, 0, 0, 0, 0,
   (1,8,0): 0, 0, 0, 0, 0, 0, 0,
   (1,9,0): 0, 0, 0, 0, 0, 0, 0,
   (1,10,0): 0, 0, 0, 0, 0, 0, 0,
   (1,11,0): 0, 0, 0, 0, 0, 0, 0,

I've tried versions 1.8.5 through 1.8.9, and they are all wrong, but in
different ways!

Here is the output from 1.8.5-patch1, run on the same file on Win32:

   DATASET "trajectory" {
      DATATYPE H5T_IEEE_F64LE
      DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
H5S_UNLIMITED ) } DATA {
      (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0,
8, (0,0,6): 0,
      (0,1,0): 0, 0, 0, 0, 0, 0, 0,
      (0,2,0): 0, 0, 1.32116e-308, 1.38263e-304, 0, 0, 0,
      (0,3,0): 0, 0, 0, 0, 0, 0, 0,
      (0,4,0): 0, 0, 0, 0, 0, 0, 0,
      (0,5,0): 0, 0, 0, 0, 0, 0, 0,
      (0,6,0): 0, 0, 0, 0, 0, 0, 0,
      (0,7,0): 0, 0, 0, 0, 0, 0, 0,
      (0,8,0): 0, 0, 0, 0, 0, 0, 0,
      (0,9,0): 0, 0, 0, 0, 0, 0, 0,
      (0,10,0): 0, 0, 1.32228e-308, 1.38263e-304, 0, 0, 0,
      (0,11,0): 0, 0, 0, 0, 0, 0, 0,

This is an anomalous case for us, as it only appears on this one file, out
of 6200+ in our test suite. We can work around it for now, but it would be
great to know what it is we are doing when we create this file that is
causing the problem. It's small enough (144K) that I figured I'd just
attached it so that anyone who wants to look at the file may do so.

Thanks!
Dave

> I'm getting some output from h5diff which seems wrong.
>
> [dws@oxygen02 history]$ /usr/local/contrib/hdf5-1.8.8-par/bin/h5diff -v -p
> 1e-30 ptclTrackTagFromAbsorber1s_HistoryOld.h5
> /scr1_oxygen/dws/jenkins/workspace/vorpalall-build/qarnodes/oxygen/vptest
> s/vpresults-oxygen/history/ptclTrackTagFromAbsorber1s/ptclTrackTagFromAbso
> rber1s_HistoryOld.h5 /trajectory dataset: </trajectory> and </trajectory>
> size: [339x7x6] [339x7x6]
> position trajectory trajectory difference relative
> ------------------------------------------------------------------------
> [ 0 1 0 ] 1.678294506e+243 0 1.678294506e+243 1
> [ 0 1 1 ] 9.047791808e+164 0 9.047791808e+164 1
> [ 0 1 2 ] 4.78541875e+175 0 4.78541875e+175 1
> [ 0 1 3 ] 4.440349806e-38 0 4.440349806e-38 1
> [ 0 1 4 ] 4.545801862e+97 0 4.545801862e+97 1
> [ 0 1 5 ] 8.535925984e-307 0 8.535925984e-307 1
> [ 0 2 0 ] 3.952525167e-322 0 3.952525167e-322 1
> [ 0 2 1 ] 3.119382268e-319 0 3.119382268e-319 1
> [ 0 2 2 ] 9.698666729e-317 0 9.698666729e-317 1
> [ 0 2 3 ] 6.920039978e-310 0 6.920039978e-310 1
> [ 0 2 4 ] 4.193339982e+228 0 4.193339982e+228 1
> [ 0 2 5 ] 1.129002081e+277 0 1.129002081e+277 1
> [ 0 3 0 ] 1.355073239e+248 0 1.355073239e+248 1
> [ 0 3 2 ] 3.162020133e-322 0 3.162020133e-322 1
> [ 0 3 3 ] 3.162020133e-322 0 3.162020133e-322 1
> [ 0 3 5 ] 2.60988666e+180 0 2.60988666e+180 1
> [ 0 4 0 ] 4.193339982e+228 0 4.193339982e+228 1
> [ 0 4 1 ] 1.129002081e+277 0 1.129002081e+277 1
> [ 0 4 2 ] 1.355073239e+248 0 1.355073239e+248 1
> [ 0 4 5 ] 3.113058228e-319 0 3.113058228e-319 1
> [ 0 5 0 ] 4.824065618e+228 0 4.824065618e+228 1
> [ 0 5 1 ] 2.60988666e+180 0 2.60988666e+180 1
> [ 0 5 2 ] 4.193339982e+228 0 4.193339982e+228 1
> [ 0 5 3 ] 1.129002081e+277 0 1.129002081e+277 1
> [ 0 5 4 ] 1.355073239e+248 0 1.355073239e+248 1
>
> The diffs says that element (0,1,0) of the /trajectory dataset is
> 1.678294506e+243 (in bold above), but when I use h5dump to look at the
> dataset, I see 0s:
>
> ...
>
> DATASET "trajectory" {
>
> DATATYPE H5T_IEEE_F64LE
> DATASPACE SIMPLE { ( 339, 7, 6 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
> H5S_UNLIMITED ) } DATA {
> (0,0,0): 1.09772e-05, 1.007e+12, 0, 0, 8, 0,
> (0,1,0): 0, 0, 0, 0, 0, 0,
> (0,2,0): 0, 0, 0, 0, 0, 0,
> (0,3,0): 0, 0, 0, 0, 0, 0,
> (0,4,0): 0, 0, 0, 0, 0, 0,
> (0,5,0): 0, 0, 0, 0, 0, 0,
> (0,6,0): 0, 0, 0, 0, 0, 0,
> (1,0,0): 1.10752e-05, 1.00993e+12, 0, 0, 8, 0,
> (1,1,0): 0, 0, 0, 0, 0, 0,
> (1,2,0): 0, 0, 0, 0, 0, 0,
> (1,3,0): 0, 0, 0, 0, 0, 0,
> (1,4,0): 0, 0, 0, 0, 0, 0,
> (1,5,0): 0, 0, 0, 0, 0, 0,
> (1,6,0): 0, 0, 0, 0, 0, 0,
>
> ...
>
> Any ideas what's going on here. The datatype in both files is
> H5T_IEEE_F64LE, so I know this is not a datatype comparison issue.
>
> Thanks,
> Dave

7

Thanks, Allen!

If uninitialized data presents a problem for h5diff/h5dump, perhaps they could indicate the data are initialized? We'll fix our code, but it would be nice to have the h5 tools indicate this, rather than returning a random value.

Dave

···

On Nov 6, 2012, at 12:11 PM, Allen D Byrne wrote:

The issue is uninitialized data in the dataset. Only the major
index(1,0,0-2,0,0-3,0,0-...) has data.

H5dump uses malloc to create the read buffer and the initial state of that
buffer is platform dependent. By changing this to a calloc call, the issue is
eliminated.

The question becomes how should h5dump behave?

Allen

On Monday, October 29, 2012 12:18:47 PM Dave Wade-Stein wrote:

So it appears this problem with h5diff extends to h5dump as well.

I have an h5 file, which, when h5dump'ed on Win32 and Linux, produces
different output.

Incorrect output is shown in bold.

Linux:
[dws@oxygen02 history]$ h5dump trajectory
ptclTrackTagFromAbsorber2s_History.h5 ...
  DATASET "trajectory" {
     DATATYPE H5T_IEEE_F64LE
     DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
H5S_UNLIMITED ) } DATA {
     (0,0,0): 1.06204e-05, -4.79586e-07, 1.00833e+12, 1.87949e+08, 0, 8, 0,
(0,1,0): 0, 0, 0, 0, 0, 0, 0,
     (0,2,0): 0, 0, 0, 0, 0, 0, 0,
     (0,3,0): 0, 0, 0, 0, 0, 0, 0,
     (0,4,0): 0, 0, 0, 0, 0, 0, 0,
     (0,5,0): 0, 0, 0, 0, 0, 0, 0,
     (0,6,0): 0, 0, 0, 0, 0, 0, 0,
     (0,7,0): 0, 0, 0, 0, 0, 0, 0,
     (0,8,0): 0, 0, 0, 0, 0, 0, 0,
     (0,9,0): 0, 0, 0, 0, 0, 0, 0,
     (0,10,0): 0, 0, 0, 0, 0, 0, 0,
     (0,11,0): 0, 0, 0, 0, 0, 0, 0,
     (1,0,0): 1.07179e-05, -4.79568e-07, 1.01124e+12, 1.89595e+08, 0, 8, 0,
(1,1,0): 0, 0, 0, 0, 0, 0, 0,
     (1,2,0): 0, 0, 0, 0, 0, 0, 0,
     (1,3,0): 0, 0, 0, 0, 0, 0, 0,
     (1,4,0): 0, 0, 0, 0, 0, 0, 0,
     (1,5,0): 0, 0, 0, 0, 0, 0, 0,
     (1,6,0): 0, 0, 0, 0, 0, 0, 0,
     (1,7,0): 0, 0, 0, 0, 0, 0, 0,
     (1,8,0): 0, 0, 0, 0, 0, 0, 0,
     (1,9,0): 0, 0, 0, 0, 0, 0, 0,
     (1,10,0): 0, 0, 0, 0, 0, 0, 0,
     (1,11,0): 0, 0, 0, 0, 0, 0, 0,
...

Win32:
...
  DATASET "trajectory" {
     DATATYPE H5T_IEEE_F64LE
     DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
H5S_UNLIMITED ) } DATA {
     (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0,
8, (0,0,6): 0,
     (0,1,0): 1.71412e+214, 1.73429e+185, 1.75754e+136, 3.44266e+218,
     (0,1,4): 5.73569e+169, 0, 1.34682e-308,
     (0,2,0): 1.91002e-308, 0, 0, 0, 0, 0, 0,
     (0,3,0): 0, 0, 0, 0, 0, 0, 0,
     (0,4,0): 0, 0, 0, 0, 0, 0, 0,
     (0,5,0): 0, 0, 0, 0, 0, 0, 0,
     (0,6,0): 0, 0, 0, 0, 0, 0, 0,
     (0,7,0): 0, 0, 0, 0, 0, 0, 0,
     (0,8,0): 0, 0, 0, 0, 0, 0, 0,
     (0,9,0): 0, 0, 0, 0, 0, 0, 0,
     (0,10,0): 0, 0, 0, 0, 0, 0, 0,
     (0,11,0): 0, 0, 0, 0, 0, 0, 0,
...

Oddly, h5dump produces still different output with the -d option. Here is
the output of h5dump -d trajectory, run on Win32:

HDF5 "ptclTrackTagFromAbsorber2s_History.h5" {
DATASET "trajectory" {
  DATATYPE H5T_IEEE_F64LE
  DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
H5S_UNLIMITED ) } DATA {
  (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0, 8,
0, (0,1,0): 0, 0, 0, 0, 0, 0, 0,
  (0,2,0): 0, 0, 0, 0, 0, 0, 0,
  (0,3,0): 0, 0, 0, 0, 0, 0, 0,
  (0,4,0): 0, 0, 0, 0, 0, 0, 0,
  (0,5,0): 0, 0, 0, 0, 0, 0, 0,
  (0,6,0): 0, 0, 0, 0, 0, 0, 0,
  (0,7,0): 0, 0, 0, 0, 0, 0, 0,
  (0,8,0): 0, 0, 0, 0, 0, 0, 0,
  (0,9,0): 0, 0, 0, 0, 0, 0, 0,
  (0,10,0): 0, 0, 0, 0, 0, 0, 0,
  (0,11,0): 0, 0, 0, 0, 0, 0, 0,
  (1,0,0): 1.07179e-005, -4.79568e-007, 1.01124e+012, 1.89595e+008, 0, 8,
0, (1,1,0): 0, 0, 0, 0, 0, 0, 0,
  (1,2,0): 0, 0, 0, 0, 0, 0, 0,
  (1,3,0): 1.37758e-308, 8.34708e-309, 4.17213e-308, 1.043e-308, 0, 0, 0,
  (1,4,0): 0, 0, 0, 0, 0, 0, 0,
  (1,5,0): 0, 0, 0, 0, 0, 0, 0,
  (1,6,0): 0, 0, 0, 0, 0, 0, 0,
  (1,7,0): 0, 0, 0, 0, 0, 0, 0,
  (1,8,0): 0, 0, 0, 0, 0, 0, 0,
  (1,9,0): 0, 0, 0, 0, 0, 0, 0,
  (1,10,0): 0, 0, 0, 0, 0, 0, 0,
  (1,11,0): 0, 0, 0, 0, 0, 0, 0,

I've tried versions 1.8.5 through 1.8.9, and they are all wrong, but in
different ways!

Here is the output from 1.8.5-patch1, run on the same file on Win32:

  DATASET "trajectory" {
     DATATYPE H5T_IEEE_F64LE
     DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
H5S_UNLIMITED ) } DATA {
     (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0,
8, (0,0,6): 0,
     (0,1,0): 0, 0, 0, 0, 0, 0, 0,
     (0,2,0): 0, 0, 1.32116e-308, 1.38263e-304, 0, 0, 0,
     (0,3,0): 0, 0, 0, 0, 0, 0, 0,
     (0,4,0): 0, 0, 0, 0, 0, 0, 0,
     (0,5,0): 0, 0, 0, 0, 0, 0, 0,
     (0,6,0): 0, 0, 0, 0, 0, 0, 0,
     (0,7,0): 0, 0, 0, 0, 0, 0, 0,
     (0,8,0): 0, 0, 0, 0, 0, 0, 0,
     (0,9,0): 0, 0, 0, 0, 0, 0, 0,
     (0,10,0): 0, 0, 1.32228e-308, 1.38263e-304, 0, 0, 0,
     (0,11,0): 0, 0, 0, 0, 0, 0, 0,

This is an anomalous case for us, as it only appears on this one file, out
of 6200+ in our test suite. We can work around it for now, but it would be
great to know what it is we are doing when we create this file that is
causing the problem. It's small enough (144K) that I figured I'd just
attached it so that anyone who wants to look at the file may do so.

Thanks!
Dave

I'm getting some output from h5diff which seems wrong.

[dws@oxygen02 history]$ /usr/local/contrib/hdf5-1.8.8-par/bin/h5diff -v -p
1e-30 ptclTrackTagFromAbsorber1s_HistoryOld.h5
/scr1_oxygen/dws/jenkins/workspace/vorpalall-build/qarnodes/oxygen/vptest
s/vpresults-oxygen/history/ptclTrackTagFromAbsorber1s/ptclTrackTagFromAbso
rber1s_HistoryOld.h5 /trajectory dataset: </trajectory> and </trajectory>
size: [339x7x6] [339x7x6]
position trajectory trajectory difference relative
------------------------------------------------------------------------
[ 0 1 0 ] 1.678294506e+243 0 1.678294506e+243 1
[ 0 1 1 ] 9.047791808e+164 0 9.047791808e+164 1
[ 0 1 2 ] 4.78541875e+175 0 4.78541875e+175 1
[ 0 1 3 ] 4.440349806e-38 0 4.440349806e-38 1
[ 0 1 4 ] 4.545801862e+97 0 4.545801862e+97 1
[ 0 1 5 ] 8.535925984e-307 0 8.535925984e-307 1
[ 0 2 0 ] 3.952525167e-322 0 3.952525167e-322 1
[ 0 2 1 ] 3.119382268e-319 0 3.119382268e-319 1
[ 0 2 2 ] 9.698666729e-317 0 9.698666729e-317 1
[ 0 2 3 ] 6.920039978e-310 0 6.920039978e-310 1
[ 0 2 4 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 2 5 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 3 0 ] 1.355073239e+248 0 1.355073239e+248 1
[ 0 3 2 ] 3.162020133e-322 0 3.162020133e-322 1
[ 0 3 3 ] 3.162020133e-322 0 3.162020133e-322 1
[ 0 3 5 ] 2.60988666e+180 0 2.60988666e+180 1
[ 0 4 0 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 4 1 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 4 2 ] 1.355073239e+248 0 1.355073239e+248 1
[ 0 4 5 ] 3.113058228e-319 0 3.113058228e-319 1
[ 0 5 0 ] 4.824065618e+228 0 4.824065618e+228 1
[ 0 5 1 ] 2.60988666e+180 0 2.60988666e+180 1
[ 0 5 2 ] 4.193339982e+228 0 4.193339982e+228 1
[ 0 5 3 ] 1.129002081e+277 0 1.129002081e+277 1
[ 0 5 4 ] 1.355073239e+248 0 1.355073239e+248 1

The diffs says that element (0,1,0) of the /trajectory dataset is
1.678294506e+243 (in bold above), but when I use h5dump to look at the
dataset, I see 0s:

...

DATASET "trajectory" {

     DATATYPE H5T_IEEE_F64LE
     DATASPACE SIMPLE { ( 339, 7, 6 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
     H5S_UNLIMITED ) } DATA {
     (0,0,0): 1.09772e-05, 1.007e+12, 0, 0, 8, 0,
     (0,1,0): 0, 0, 0, 0, 0, 0,
     (0,2,0): 0, 0, 0, 0, 0, 0,
     (0,3,0): 0, 0, 0, 0, 0, 0,
     (0,4,0): 0, 0, 0, 0, 0, 0,
     (0,5,0): 0, 0, 0, 0, 0, 0,
     (0,6,0): 0, 0, 0, 0, 0, 0,
     (1,0,0): 1.10752e-05, 1.00993e+12, 0, 0, 8, 0,
     (1,1,0): 0, 0, 0, 0, 0, 0,
     (1,2,0): 0, 0, 0, 0, 0, 0,
     (1,3,0): 0, 0, 0, 0, 0, 0,
     (1,4,0): 0, 0, 0, 0, 0, 0,
     (1,5,0): 0, 0, 0, 0, 0, 0,
     (1,6,0): 0, 0, 0, 0, 0, 0,

...

Any ideas what's going on here. The datatype in both files is
H5T_IEEE_F64LE, so I know this is not a datatype comparison issue.

Thanks,
Dave

8

The tools lead decided that h5dump (h5diff) should use calloc calls. That
solution is now in process.

Allen

···

On Tuesday, November 06, 2012 12:50:33 PM Dave Wade-Stein wrote:

Thanks, Allen!

If uninitialized data presents a problem for h5diff/h5dump, perhaps they
could indicate the data are initialized? We'll fix our code, but it would
be nice to have the h5 tools indicate this, rather than returning a random
value.

Dave

On Nov 6, 2012, at 12:11 PM, Allen D Byrne wrote:
> The issue is uninitialized data in the dataset. Only the major
> index(1,0,0-2,0,0-3,0,0-...) has data.
>
> H5dump uses malloc to create the read buffer and the initial state of that
> buffer is platform dependent. By changing this to a calloc call, the issue
> is eliminated.
>
> The question becomes how should h5dump behave?
>
> Allen
>
> On Monday, October 29, 2012 12:18:47 PM Dave Wade-Stein wrote:
>> So it appears this problem with h5diff extends to h5dump as well.
>>
>> I have an h5 file, which, when h5dump'ed on Win32 and Linux, produces
>> different output.
>>
>> Incorrect output is shown in bold.
>>
>> Linux:
>> [dws@oxygen02 history]$ h5dump trajectory
>> ptclTrackTagFromAbsorber2s_History.h5 ...
>>
>> DATASET "trajectory" {
>>
>> DATATYPE H5T_IEEE_F64LE
>> DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
>>
>> H5S_UNLIMITED ) } DATA {
>>
>> (0,0,0): 1.06204e-05, -4.79586e-07, 1.00833e+12, 1.87949e+08, 0, 8,
>> 0,
>>
>> (0,1,0): 0, 0, 0, 0, 0, 0, 0,
>>
>> (0,2,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,3,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,4,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,5,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,6,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,7,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,8,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,9,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,10,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,11,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,0,0): 1.07179e-05, -4.79568e-07, 1.01124e+12, 1.89595e+08, 0, 8,
>> 0,
>>
>> (1,1,0): 0, 0, 0, 0, 0, 0, 0,
>>
>> (1,2,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,3,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,4,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,5,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,6,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,7,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,8,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,9,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,10,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,11,0): 0, 0, 0, 0, 0, 0, 0,
>>
>> ...
>>
>> Win32:
>> ...
>>
>> DATASET "trajectory" {
>>
>> DATATYPE H5T_IEEE_F64LE
>> DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
>>
>> H5S_UNLIMITED ) } DATA {
>>
>> (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0,
>>
>> 8, (0,0,6): 0,
>>
>> (0,1,0): 1.71412e+214, 1.73429e+185, 1.75754e+136, 3.44266e+218,
>> (0,1,4): 5.73569e+169, 0, 1.34682e-308,
>> (0,2,0): 1.91002e-308, 0, 0, 0, 0, 0, 0,
>> (0,3,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,4,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,5,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,6,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,7,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,8,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,9,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,10,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,11,0): 0, 0, 0, 0, 0, 0, 0,
>>
>> ...
>>
>> Oddly, h5dump produces still different output with the -d option. Here is
>> the output of h5dump -d trajectory, run on Win32:
>>
>> HDF5 "ptclTrackTagFromAbsorber2s_History.h5" {
>> DATASET "trajectory" {
>>
>> DATATYPE H5T_IEEE_F64LE
>> DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
>>
>> H5S_UNLIMITED ) } DATA {
>>
>> (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0, 8,
>>
>> 0, (0,1,0): 0, 0, 0, 0, 0, 0, 0,
>>
>> (0,2,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,3,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,4,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,5,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,6,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,7,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,8,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,9,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,10,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,11,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,0,0): 1.07179e-005, -4.79568e-007, 1.01124e+012, 1.89595e+008, 0, 8,
>>
>> 0, (1,1,0): 0, 0, 0, 0, 0, 0, 0,
>>
>> (1,2,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,3,0): 1.37758e-308, 8.34708e-309, 4.17213e-308, 1.043e-308, 0, 0, 0,
>> (1,4,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,5,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,6,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,7,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,8,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,9,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,10,0): 0, 0, 0, 0, 0, 0, 0,
>> (1,11,0): 0, 0, 0, 0, 0, 0, 0,
>>
>> I've tried versions 1.8.5 through 1.8.9, and they are all wrong, but in
>> different ways!
>>
>> Here is the output from 1.8.5-patch1, run on the same file on Win32:
>> DATASET "trajectory" {
>>
>> DATATYPE H5T_IEEE_F64LE
>> DATASPACE SIMPLE { ( 218, 12, 7 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
>>
>> H5S_UNLIMITED ) } DATA {
>>
>> (0,0,0): 1.06204e-005, -4.79586e-007, 1.00833e+012, 1.87949e+008, 0,
>>
>> 8, (0,0,6): 0,
>>
>> (0,1,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,2,0): 0, 0, 1.32116e-308, 1.38263e-304, 0, 0, 0,
>> (0,3,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,4,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,5,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,6,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,7,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,8,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,9,0): 0, 0, 0, 0, 0, 0, 0,
>> (0,10,0): 0, 0, 1.32228e-308, 1.38263e-304, 0, 0, 0,
>> (0,11,0): 0, 0, 0, 0, 0, 0, 0,
>>
>> This is an anomalous case for us, as it only appears on this one file,
>> out
>> of 6200+ in our test suite. We can work around it for now, but it would
>> be
>> great to know what it is we are doing when we create this file that is
>> causing the problem. It's small enough (144K) that I figured I'd just
>> attached it so that anyone who wants to look at the file may do so.
>>
>> Thanks!
>> Dave
>>
>>> I'm getting some output from h5diff which seems wrong.
>>>
>>> [dws@oxygen02 history]$ /usr/local/contrib/hdf5-1.8.8-par/bin/h5diff -v
>>> -p
>>> 1e-30 ptclTrackTagFromAbsorber1s_HistoryOld.h5
>>> /scr1_oxygen/dws/jenkins/workspace/vorpalall-build/qarnodes/oxygen/vptes
>>> t
>>> s/vpresults-oxygen/history/ptclTrackTagFromAbsorber1s/ptclTrackTagFromAb
>>> so
>>> rber1s_HistoryOld.h5 /trajectory dataset: </trajectory> and
>>> </trajectory>
>>> size: [339x7x6] [339x7x6]
>>> position trajectory trajectory difference relative
>>> ------------------------------------------------------------------------
>>> [ 0 1 0 ] 1.678294506e+243 0 1.678294506e+243 1
>>> [ 0 1 1 ] 9.047791808e+164 0 9.047791808e+164 1
>>> [ 0 1 2 ] 4.78541875e+175 0 4.78541875e+175 1
>>> [ 0 1 3 ] 4.440349806e-38 0 4.440349806e-38 1
>>> [ 0 1 4 ] 4.545801862e+97 0 4.545801862e+97 1
>>> [ 0 1 5 ] 8.535925984e-307 0 8.535925984e-307 1
>>> [ 0 2 0 ] 3.952525167e-322 0 3.952525167e-322 1
>>> [ 0 2 1 ] 3.119382268e-319 0 3.119382268e-319 1
>>> [ 0 2 2 ] 9.698666729e-317 0 9.698666729e-317 1
>>> [ 0 2 3 ] 6.920039978e-310 0 6.920039978e-310 1
>>> [ 0 2 4 ] 4.193339982e+228 0 4.193339982e+228 1
>>> [ 0 2 5 ] 1.129002081e+277 0 1.129002081e+277 1
>>> [ 0 3 0 ] 1.355073239e+248 0 1.355073239e+248 1
>>> [ 0 3 2 ] 3.162020133e-322 0 3.162020133e-322 1
>>> [ 0 3 3 ] 3.162020133e-322 0 3.162020133e-322 1
>>> [ 0 3 5 ] 2.60988666e+180 0 2.60988666e+180 1
>>> [ 0 4 0 ] 4.193339982e+228 0 4.193339982e+228 1
>>> [ 0 4 1 ] 1.129002081e+277 0 1.129002081e+277 1
>>> [ 0 4 2 ] 1.355073239e+248 0 1.355073239e+248 1
>>> [ 0 4 5 ] 3.113058228e-319 0 3.113058228e-319 1
>>> [ 0 5 0 ] 4.824065618e+228 0 4.824065618e+228 1
>>> [ 0 5 1 ] 2.60988666e+180 0 2.60988666e+180 1
>>> [ 0 5 2 ] 4.193339982e+228 0 4.193339982e+228 1
>>> [ 0 5 3 ] 1.129002081e+277 0 1.129002081e+277 1
>>> [ 0 5 4 ] 1.355073239e+248 0 1.355073239e+248 1
>>>
>>> The diffs says that element (0,1,0) of the /trajectory dataset is
>>> 1.678294506e+243 (in bold above), but when I use h5dump to look at the
>>> dataset, I see 0s:
>>>
>>> ...
>>>
>>> DATASET "trajectory" {
>>>
>>> DATATYPE H5T_IEEE_F64LE
>>> DATASPACE SIMPLE { ( 339, 7, 6 ) / ( H5S_UNLIMITED, H5S_UNLIMITED,
>>> H5S_UNLIMITED ) } DATA {
>>> (0,0,0): 1.09772e-05, 1.007e+12, 0, 0, 8, 0,
>>> (0,1,0): 0, 0, 0, 0, 0, 0,
>>> (0,2,0): 0, 0, 0, 0, 0, 0,
>>> (0,3,0): 0, 0, 0, 0, 0, 0,
>>> (0,4,0): 0, 0, 0, 0, 0, 0,
>>> (0,5,0): 0, 0, 0, 0, 0, 0,
>>> (0,6,0): 0, 0, 0, 0, 0, 0,
>>> (1,0,0): 1.10752e-05, 1.00993e+12, 0, 0, 8, 0,
>>> (1,1,0): 0, 0, 0, 0, 0, 0,
>>> (1,2,0): 0, 0, 0, 0, 0, 0,
>>> (1,3,0): 0, 0, 0, 0, 0, 0,
>>> (1,4,0): 0, 0, 0, 0, 0, 0,
>>> (1,5,0): 0, 0, 0, 0, 0, 0,
>>> (1,6,0): 0, 0, 0, 0, 0, 0,
>>>
>>> ...
>>>
>>> Any ideas what's going on here. The datatype in both files is
>>> H5T_IEEE_F64LE, so I know this is not a datatype comparison issue.
>>>
>>> Thanks,
>>> Dave