I am running a source code on Ubuntu 14.04, 64 bit. And on 32 bit 10.04 Using intel Fortran. The tests differ substantially from each other.
next i tried using gfortarn compiler. same thing observed.
the intel compiler and gfortran on 14.04 n 64 bits machines were identical and so were on 10.04 machine
Can some one help me in understanding the issue.
How can results match and which one should we rely upon.
thanks in anticipation
It all depends on what your program does. Programs with bugs can sometimes give correct results. For this reason, bugs may remain hidden for a long time.
If you wish, you can post the program source code here and someone may respond.
We coexist with bugs all the time, whether in software, in the air or in our guts.
Whether there is a bug in a library that affects the running and the results of your program is what is more important. That question can only be answered if you provide a small test case that displays evidence of a bug.
The older machine may have had a compiler that generated x87 instructions for real number arithmetic. With 80-bit registers, the x87 provides more precision than the 64-bit SSE registers, so if your algorithm is sensitive to insufficient precision, you need to rethink the calculation.