[netcdfgroup] netCDF4 file bit-for-bit reproducibility?

[Quincey Koziol]

Hi Ed,

On Jan 8, 2008, at 9:39 AM, Ed Hartnett wrote:

> Charlie Zender <zender at uci.edu> writes:
>
>> Hi and happy new year to the netCDF group,
>>
>> Will netCDF4 make bit-for-bit (BFB) reproducible files?
>> In other words will running a determinstic model twice on the same
>> machine (without changing compiler, netCDF library, etc.) produce
>> identical output files?
>> I thought that netCDF4 would, like netCDF3, use deterministic
>> algorithms without any date-stamps within the file, and thus produce
>> BFB files. Apparently I am wrong.
>>
>> To determine whether files were BFB I checked their SHA1 sums.
>> netCDF3 produces BFB files and netCDF4 (snapshot) does not (below).
>> BFB output files make models easier to debug, so it would be helpful
>> if netCDF4 continued the netCDF3 BFB "tradition". Is this possible?
>> Am I missing something? Is HDF5 the culprit?
>>
>> Thanks,
>> Charlie
>>
>
> Howdy Charlie!
>
> I is my understanding that HDF5 does not guarantee the bit-or-bit
> reproducibility of their files. I have forwarded this to them to
> confirm this...

	If you turn off the create/modify/access time tracking for objects  
created (with the H5Pset_obj_track_times() routine), everything should  
be bit-for-bit reproducible.  Coincidentally, it makes accessing those  
objects faster and the size of their metadata smaller also.  You do  
lose the ability to know when the object was created/modified/accessed.

	Quincey


> I'm afraid there is nothing I can do at the netCDF-4 level to change
> that.
>
> Thanks,
>
> Ed
> -- 
> Ed Hartnett  -- ed at unidata.ucar.edu
>