- 新着としてマーク
- ブックマーク
- 購読
- ミュート
- RSS フィードを購読する
- ハイライト
- 印刷
- 不適切なコンテンツを報告
Hello,
I use ifort in conjunction with -fpp flag since there is some preprocessing in the code. The preprocessing directives are all of #ifndef #endif type, to include or exclude pieces of code according to my desires.
I have noticed that with -fpp, each time I compile the code I get a different executable. This was checked by using cmp compand on Linux.
Without preprocessing, the produced binaries are identical.
To be completely precise,I compare, of course, preprocessed to preprocessed instances, and the same for non-preprocessed.
What could be the reason for the differences, when the code, optimisation et al. is always the same?
The compiler version is 10.1, and it works on a Linux cluster (i don't have all the specific details for the platform, if requested i'll bother the cluster support guys)
I use ifort in conjunction with -fpp flag since there is some preprocessing in the code. The preprocessing directives are all of #ifndef #endif type, to include or exclude pieces of code according to my desires.
I have noticed that with -fpp, each time I compile the code I get a different executable. This was checked by using cmp compand on Linux.
Without preprocessing, the produced binaries are identical.
To be completely precise,I compare, of course, preprocessed to preprocessed instances, and the same for non-preprocessed.
What could be the reason for the differences, when the code, optimisation et al. is always the same?
The compiler version is 10.1, and it works on a Linux cluster (i don't have all the specific details for the platform, if requested i'll bother the cluster support guys)
コピーされたリンク
2 返答（返信）
- 新着としてマーク
- ブックマーク
- 購読
- ミュート
- RSS フィードを購読する
- ハイライト
- 印刷
- 不適切なコンテンツを報告
Quoting - smorifort
Hello,
I use ifort in conjunction with -fpp flag since there is some preprocessing in the code. The preprocessing directives are all of #ifndef #endif type, to include or exclude pieces of code according to my desires.
I have noticed that with -fpp, each time I compile the code I get a different executable. This was checked by using cmp compand on Linux.
Without preprocessing, the produced binaries are identical.
To be completely precise,I compare, of course, preprocessed to preprocessed instances, and the same for non-preprocessed.
What could be the reason for the differences, when the code, optimisation et al. is always the same?
The compiler version is 10.1, and it works on a Linux cluster (i don't have all the specific details for the platform, if requested i'll bother the cluster support guys)
I use ifort in conjunction with -fpp flag since there is some preprocessing in the code. The preprocessing directives are all of #ifndef #endif type, to include or exclude pieces of code according to my desires.
I have noticed that with -fpp, each time I compile the code I get a different executable. This was checked by using cmp compand on Linux.
Without preprocessing, the produced binaries are identical.
To be completely precise,I compare, of course, preprocessed to preprocessed instances, and the same for non-preprocessed.
What could be the reason for the differences, when the code, optimisation et al. is always the same?
The compiler version is 10.1, and it works on a Linux cluster (i don't have all the specific details for the platform, if requested i'll bother the cluster support guys)
I have no idea. Dynamic library names are included in the header sections, if your LD_LIBRARY_PATH changes OR if the sys admins upgraded some system libraries or gcc, this could account for the changes.
You could try using od -ax on both and find the differences.
ron
- 新着としてマーク
- ブックマーク
- 購読
- ミュート
- RSS フィードを購読する
- ハイライト
- 印刷
- 不適切なコンテンツを報告
When you preprocess to .f90 source files, save files under different name/place, then preprocess to files again to .f90 source files, then compare the preprocessed files, do the files compare as equal?
What are your options relating to preprocessing?
Jim Dempsey
