suspicous compiler flag combination

[AronRoland]

Hi everbody,

I am working on some operational forecasting model and I found that the following flags have been utilized for the intel compiler, which i found a bit weird.

Maybe somebody is will to comment on that.

-no-fma -ip -g -traceback -i4 -real-size 32 -i8 -ilp64  -fp-model precise -assume byterecl  -fno-alias -fno-fnalias

I am particulary wondering about this part " -i4 -real-size 32 -i8 -ilp64 "

Thanks in advance

Cheers

Aron

[Ron_Green]

left to right, -i4 .. -i8 will set Integer size to 8bytes.  Right-most option takes precendence.

Confirm what you are getting by using option -dryrun to dump all the compiler settings used for the compilation.  Just add -dryrun to the compilation line.

[AronRoland]

Hi Ron,

let me thank u 1st for the new flag -dryrun that I was not aware of, very nice and for clarification of the order of the flags. From my point of view i think it is not really appropriate manipulating variable precision by compiler flags, I found that very weird. Especially, if somebody is coding something by purpose, like e.g. me, putting 1 byte integer if only 0 and 1 are option than this kind of possible alignment as well as the memory overhead get introduced by those flags, however, who cares ...

Many many thanks for your fast and nice answer!

Cheers

Aron