[Users] Intel Compiler Issues

[Erik Schnetter]

Tim

This function is a timing function. The assembler statements has two
outputs, namely the variables a and d. Apparently the compiler doesn't
understand that.

You can manually initialise a and d before the assembler statement to
circumvent this. This is a compiler error; consider reporting it to
Intel.

-erik

On Tue, Aug 9, 2011 at 11:47 AM, Tim Lemon <tim.lemon at soton.ac.uk> wrote:
> Hello,
>
> I'm currently trying to use the Southampton supercomputer to run my
> simulations using the Einstein Toolkit.
>
> I'm using the latest version of carpet (Mercurial) and the I've loaded the
> Intel 11.1 compiler, but I am getting the following error.
>
> "Run-Time Check Failure: The variable 'd' is being used without being
> initialized"
>
> This can be traced back to a function in CarpetLib (cycle.h),
>
> #if (defined(__GNUC__) || defined(__ICC) || defined(__SUNPRO_C)) &&
> defined(__x86_64__)  && !defined(HAVE_TICK_COUNTER)
> typedef unsigned long long ticks;
>
> static __inline__ ticks getticks(void)
> {
>      unsigned a, d;
>      asm volatile("rdtsc" : "=a" (a), "=d" (d));
>      return ((ticks)a) | (((ticks)d) << 32);
> }
>
>
> Any suggestions on how to fix this problem?
>
> --
> Kind Regards
>
> Tim lemon
>
> _______________________________________________
> Users mailing list
> Users at einsteintoolkit.org
> http://lists.einsteintoolkit.org/mailman/listinfo/users
>
>



-- 
Erik Schnetter <schnetter at cct.lsu.edu>   http://www.cct.lsu.edu/~eschnett/