Unusual Behavior

Accelerated Computing HPC Compilers Legacy PGI Compilers
1

Hi,

I have used gfortran and Intel Fortran in the past. I recently switched to PGI Fortran. I’m noticing an odd behavior from the compiler. Whenever a stop command is issued, this is what is output to screen:

*** PROGRAM STOPPED; THRUST LESS THAN ZERO ***
Warning: ieee_invalid is signaling
Warning: ieee_denorm is signaling
Warning: ieee_divide_by_zero is signaling
Warning: ieee_inexact is signaling
FORTRAN STOP

I’ve never seen this before from the other compilers. Is there any way to have it just show my output message rather than whatever the compiler is showing additionally. That’s how the other compilers worked. I think the additional messages are confusing. The stop was issued intentionally. It wasn’t any of the things the compiler is saying.

2

Hi prop_design,

As of the F2003 standard, STOP is required to print out any exceptions that may have occurred. You can disable this message by setting the environment “NO_STOP_MESSAGE=1”.

See: PGI Compiler User's Guide Version 19.10 for x86 and NVIDIA Processors

Hope this helps,
Mat

3

thanks,

i’m confused by this:

Warning: ieee_invalid is signaling
Warning: ieee_denorm is signaling
Warning: ieee_divide_by_zero is signaling
Warning: ieee_inexact is signaling

as far as i know nothing like that is happening. i put the stop in there when different variables i calculate are less than zero.

thanks for showing how to turn it off. that seems like that is something for just me though. i distribute the exe files to others. so i imagine they will see these messages if a stop occurs, is that right?

anthony

4

is there a way to turn off these strange messages when a STOP command is issued. i just want to show the message that i output, not the ones that the compiler adds in. i am referring to the *exe files i distribute. i don’t want my end users seeing a bunch of weird nonsense. i already post a message as to why the program has been stopped. if this isn’t possible, could it be a new feature in a future community edition release of pgfortran.

thanks

5

Hi prop_design,

I added an RFE (TPR #28152) to see if we can add a command line switch to suppress the F2003 warnings regarding exceptions.

-Mat

6

thanks that would be great