implied loop for write

[Yu__Wenbin]

I have a code to write a large array. 

WRITE(EIN,'(1X,i8,2ES20.10)') (i,coord(i,:),i=1,nnode)

it stops at i=63800. Do you know why? Thanks a lot for your help in advance!

[jimdempseyatthecove]

>>I think the issue is that coord(i,:) creates a stack temp that is not removed on each iteration through the loop.

I presume that this is due to the newer generation of programmers (compiler writers) favoring recursive algorithms as opposed to iterative algorithms....
... the code passes simple tests.... but fails in the users hands.

Hmmm.... wouldn't it be ironic if the cause was old code contained a GOTO that could be removed by recursively calling the procedure....

;)

Jim Dempsey

[Steve_Lionel]

No recursion here - just creation of a stack temp in a loop without cutting it back until the end of the statement.

[Yu__Wenbin]

Steve, your suggested solution works. 

[Steve_Lionel]

Glad to hear it. 

[andrew_4619]

Has a bug report been filed?

[andrew_4619]

Has a bug report been filed?

[Steve_Lionel]

I submitted this as issue 04460165.

[andrew_4619]

Thanks Steve, I would have done that in no one had because that needs fixing IMO

[Ron_Green]

Intel bug ID is CMPLRIL0-32365

[Yu__Wenbin]

For windows add /link /stack:100000000 to the end of the ifort command works. However, for Linux, I can use command ulimit -s unlimited, before executing the code to resolve this issue. If I want to resolve it during compiling and linking, what command should I use? 

[Yu__Wenbin]

I tried to compile using -heap-arrays option, a code runs for 6 seconds using ulimit -s unlimited now runs for 10 seconds. 

[Steve_Lionel]

Linux doesn't allow you to set this during the build (and Windows doesn't allow you to set it during execution.)

[Steve_Lionel]

This bug got fixed a while ago - the program now completes normally with the 2021.4 compiler, but I don't know how far back the fix goes. The compiler now properly cuts back the stack on each iteration (the correct fix.)