My Community

I am working on an initial build of VASP 5.3.3. All files compiled and linked with the PGI compiler, and I've successfully (it seems) made an executable. When I run a simple problem, however, I get the following error.

vasp.5.3.3 18Dez12 (build Mar 9 2013 12:13:51) complex
POSCAR found : 1 types and 2 ions
PGFIO-F-207/OPEN/unit=1/file is already connected to another unit.
File name = INCAR
In source file charge.f, at line number 1365[/color]

I am looking for ideas of how to address this issue.

Thanks much

Hi,
I am working on VASP 5.3.3. Files were compiled using PGI 13.6 for Ubuntu 12.04.
When I run a problem, I have the same error described above:
"PGFIO-F-207/OPEN/unit=1/file is already connected to another unit."

Any help would be highly appreciated.
Liliana

Just ran into this problem as well. Posting my particular tracking-down-and-solution process, in case it helps.

* Fortran standard forbids having two different units attached to the same open file simultaneously, hence the error.
* Failed OPEN in charge.F within subroutine GGA_COMPAT_MODE(IU5, ...).
* main.F, ~line 526 calls this routine. Line 340 defines IO%IU5 as 7; however, GGA_COMPAT_MODE gets called with value 1.
* Discovered that previous call to READER changed the value.
* Tracked the value change to call to RDATAB around line 1280 in reader.F.
* Discovered that RDATAB in the vasp support library has a single test to reassign the IU argument (IU5, in this call) if the incoming argument is <0 or >99. Perplexed, as it should be 7.
* When rebuilding with write statement, discovered ***I ACCIDENTALLY BUILT -i8***, whereas main VASP was 32-bit int. So, IU5 from VASP was being interpreted as garbage, and getting re-assigned to the next free unit (= unit 1).
* Took out -i8 for the library build, problem solved. Moral, make sure my integer widths are consistent between the libdmy.a build and the main VASP executable builds!