Issues found after updating to IVF 2011.9.3000

[osrondon]

Hi

I recently updated to IVF 2011.9.3000 from the previous version and I got some issues I did not get before when debugging/coding:

1) It appears IVF is not allocating allocatable arrays (previously was working fine). When debugging the program I get the message "Too many array subscripts" when trying to get an array value, for instance V(1). I did not get any error when allocating V. Also I got the message "undefined pointer/array".

2) Within a subroutine, it appears IVF is not reading/recognising the declaration part. It jumps straight from the subroutine name to the main body of the subroutine.

Any help or comments to get this issue solved is highly appreciated.

Thanks

[Steven_L_Intel1]

Would you please attach a ZIP of project(s) that show these behaviors? Your theory as to what is happening is likely mistaken (at least the theory doesn't match the symptoms), but I'd like to see what you are seeing.

[Steven_L_Intel1]

Thanks for the copy of your program.

I do see that there are some coding errors regarding the functions HERMITE, CUMNORM and r4_normal_01_cdf_inverse. The actual functions are declared as REAL*8, but there is no similar declaration where they are called in several subroutines. This will cause unpredictable behavior.

As for debugging, the declaration section often does not contain any executable code, so the debugger will skip over it.

I do note that the debugger is not seeing arrays Zv and Yv as not being allocated when they were allocated. We'll look into that further.

[osrondon]

Hi Steve,

Thanks for pointing out the coding errors. I did notice those when calling HERMITE but I did notsee a similar issue when calling CUMNORM or r4_normal_01_cdf.inverse.

Can you please indicate a couple of lines where you found the coding errors ?

Thanking you in advance

[Steven_L_Intel1]

Pretty much every place those functions were called, they were not declared as REAL*8 (or DOUBLE PRECISION) in the caller.

The programs ran to completion for me - do I understand correctly that the problems you encountered were only with the debugger?

[osrondon]

Hi Steve,

Thanks.

Yes, I have only problem with the debugger. (showing an array is undefined and not showing its entries when the array is actually allocated and with values for each entry)

[commbank]

Hi Steve,

I have the same problem even after installing IVF Update 2011.10.325, downloaded and installed on 21 May 2012.

Once allocated, the debugger should read {....}

but instead it reads "Undefined pointer/array"

Normally I would have isolated the error in a simple test example and send that complete simple example.

BUT, my simple example actually WORKED!

Of course I cannot guarantee complete similarity between my real application and simple example but it is about 90-95% similar.

Can you please keep us up-to-date about this problem, because it certainly has not been fixed in 2011.10.325

Note: I think this is the same issue with a few major version back (2007) - seehttp://software.intel.com/en-us/forums/showthread.php?t=54142

If that version 9.xxx was fixed, perhaps that can be used as a reference to fix this.

Thanks for your help.

Clinton

[Steven_L_Intel1]

I will need to see a real program that demonstrates the problem in order to have it investigated.