The problem is strange for me, I have fresh installed openSUSE 11.3 after that my fortran codes using gfortran compiler gives different output. The same codes gives correct values in windows but I use linux for programming.
Please help me to solve the problem.
Versions??? Maybe you need a parameter to generate older code patterns…
The version is
GCC-fortran 4.5-4.2
GCC45-fortran 4.5.0_20100604-1.12
library installed is
libgfortran45 4.5.0_20100604-1.12
It is too strange for me, checked many times. Codes complied normally, but the output is not correct, these are different in openSUSE and Windows.
When you say different, what do you mean. Binary different or functionally different? A Windows binary would be different then a Linux binary even if the code is the same.
The output from same code is given below:
Linux Windows
3.5089999999999999 -1.25982602363416422E-309 3.5089999999999999 -108070.92167079351
3.5190000000000001 -1.24257611352839618E-309 3.5190000000000001 -106591.18268324799
3.5289999999999999 -1.22465577795257682E-309 3.5289999999999999 -105053.93297895517
3.5390000000000001 -1.20618548460569455E-309 3.5390000000000001 -103469.50656783844
3.5489999999999999 -1.18727252757412510E-309 3.5489999999999999 -101847.10739613992
3.5590000000000002 -1.16801238080656836E-309 3.5590000000000002 -100194.92544843904
3.5690000000000000 -1.14848991119454019E-309 3.5690000000000000 -98520.240814039367
3.5790000000000002 -1.12878046644262293E-309 3.5790000000000002 -96829.516999239175
3.5890000000000000 -1.10895085068764837E-309 3.5890000000000000 -95128.484626300065
3.5990000000000002 -1.08906020006982944E-309 3.5990000000000002 -93422.216535153013
3.6090000000000000 -1.06916076851196189E-309 3.6090000000000000 -91715.195193177220
3.6190000000000002 -1.04929863327328371E-309 3.6190000000000002 -90011.373220110196
3.6290000000000000 -1.02951432868758291E-309 3.6290000000000000 -88314.227747883502
3.6389999999999998 -1.00984341541625229E-309 3.6389999999999998 -86626.809257627072
3.6490000000000000 -9.90316992135295415E-310 3.6490000000000000 -84951.785467165464
3.6589999999999998 -9.70962155340568366E-310 3.6589999999999998 -83291.480781021935
3.6690000000000000 -9.51802412911110293E-310 3.6690000000000000 -81647.911760404939
3.6789999999999998 -9.32858055902346176E-310 3.6789999999999998 -80022.819022105439
3.6890000000000001 -9.14146493074249900E-310 3.6890000000000001 -78417.695931986105
3.6989999999999998 -8.95682551836649966E-310 3.6989999999999998 -76833.814420229144
3.7090000000000001 -8.77478749026466929E-310 3.7090000000000001 -75272.248211161495
3.7189999999999999 -8.59545534620667144E-310 3.7189999999999999 -73733.893729859323
3.7290000000000001 -8.41891511143513798E-310 3.7290000000000001 -72219.488920406075
3.7389999999999999 -8.24523631093044864E-310 3.7389999999999999 -70729.630186297974
3.7490000000000001 -8.07447374778421305E-310 3.7490000000000001 -69264.787641707779
3.7589999999999999 -7.90666910381326945E-310 3.7589999999999999 -67825.318842868786
Here first two column for Linux and last two for Windows. In openSUSE 11.2(I do not remember the gfortran version), the out put was same as in Wndows. Actually I get to know about that when I find that my results are too strange. When I did not find any reason, I checked it on windows.
But I can not use Windows, I use openSUSE. So it is really serious for me.
Please help me to solve the problem, it is indeed important for me.
victorray wrote:
>
> Please help me to solve the problem, it is indeed important for me.
>
>
From looking at the output produced by some code which nobody knows compiled
with compiler switches you do not tell us, without telling us the
architecture of the gfortran (32 or 64 bit) and which target architecture
was choosen when compiling (and yes - this has an influence because it
affects the default precission in some cases in fortran) this is just wild
guessing.
If you use an unstable algorithm that can lead to some kind of random number
generator.
Do you link against a third party library?
Is the code itself bugfree?
Did you compile with debugging symbols and did you debug it?
You tell us about a black box without details and let us speculate about its
outcome.
without telling us the architecture of the gfortran (32 or 64 bit) and which target architecture was choosen when compiling (and yes - this has an influence because it affects the default precission in some cases in fortran) this is just wild
guessing.
The compiler installed is 64 bit and I used only 64 bit in openSUSE
If you use an unstable algorithm that can lead to some kind of random number
generator.
The code is not related to the random number generator, it is a application of simple physical formula.
Do you link against a third party library?
Is the code itself bugfree?
Did you compile with debugging symbols and did you debug it?
Not using any library, it is itself bugfree, I did not bug it.
It is a simple code working fine before in openSUSE 11.2.
How it is possible that the same code gives different results in different OS rather it is not related to random number generator.
Thank you in advanced.
without telling us the architecture of the gfortran (32 or 64 bit) and which target architecture was choosen when compiling (and yes - this has an influence because it affects the default precission in some cases in fortran) this is just wild
guessing.
The compiler installed is 64 bit and I used only 64 bit in openSUSE
If you use an unstable algorithm that can lead to some kind of random number
generator.
The code is not related to the random number generator, it is a application of simple physical formula.
Do you link against a third party library?
Is the code itself bugfree?
Did you compile with debugging symbols and did you debug it?
Not using any library, it is itself bugfree, I did not bug it.
It is a simple code working fine before in openSUSE 11.2.
How it is possible that the same code gives different results in different OS rather it is not related to random number generator.
Thank you in advanced.
Until now you refuse to answer the details asked for (only one detail at a
time but not the complete details which are needed).
Every programmer running into that kind of trouble (and yes things like that
can easily happen) would first compile his code with debug symbols and run a
debbuger of his/her choice on it to see what happens.
You do not even tell us anything what YOU did to check this, to do a root
cause analysis, before posting here in all kind of subforums with the same
missing information.
I am out here.