[Mitgcm-support] curious

[mitgcm-support at dev.mitgcm.org]

running exp5, with the new version of cg2d.F & solve_for_pressure.F
(i.e. with write(0,..  CG2D iters, err =' in solve_for_pressure.F)
give different "Numbers" (differences of the order of truncation error) :

new version, 2nd iter :
 cg2d: Sum(rhs) =          -2.16493489801906E-15 
presious version, 2nd iter :
 cg2d: Sum(rhs) =          -3.49720252756924E-15
And also:
Binary files U.0000000010.001.001.data and ../res_sav/U.0000000010.001.001.data differ

But if I activate the 2 commented "write(0," in the new version of 
cg2d.F (-> as before) (lines 227,228  and lines 390,391)
then I get the same "Numbers" as before. 

Curious, no ? 

Jean-Michel