[MITgcm-devel] DFLOAT and TAF don't like each other

Jean-Michel Campin jmc at ocean.mit.edu
Wed Jan 26 21:52:45 EST 2005


Hi Patrick,

> the reason why the adjoint tests fail since yesterday
> is your change of FLOAT -> DFLOAT in external_fields_load.F
> If you don't mind, I'll put it back to FLOAT for now.
> I'll also request to Fastopt that the make
> DFLOAT work with TAF.

OK. 
I still want to give a try with g77 and this -sse option.
I will put this DFLOAT within #ifdef ALLOW_AUTODIFF_TAMC 
for now. If this happens to be a dead-end, we will go back 
to the old g77 option and get rid of the DFLOAT.

The purpose is to try to get closer agreement between different 
platforms, and it's hard to test all of them without checking-in 
the new output.txt

Is it OK like this ?

Jean-Michel



More information about the MITgcm-devel mailing list