- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I don't know whether the 6.6C3 upgrade is official yet, as
I have only seen it discussed here by Steve Lionel.
I have just tried this update, and had to revert to 6.6B because of several compiler internal error crashes when compiling a large project. Curiously enough the modules it crashed on were the same ones which caused internal errors to earlier releases of Intel Fortran 8.0 for Linux, and the workarounds I developed for the Linux compiler also function as workarounds for CVF 6.6C3 crashes.
I know these compilers are supposed to share code but this was unexpected behaviour.
In case it helps identify the problems, the crashes were associated with "use" statements which use associate other modules which use associated still other modules. The workaround is not to put "use" statements at the module level but in each subroutine individually, and to explicitly specify the variables with "only". Perhaps this is suggestive of a symbol-table overflow.
Keith Refson
I have only seen it discussed here by Steve Lionel.
I have just tried this update, and had to revert to 6.6B because of several compiler internal error crashes when compiling a large project. Curiously enough the modules it crashed on were the same ones which caused internal errors to earlier releases of Intel Fortran 8.0 for Linux, and the workarounds I developed for the Linux compiler also function as workarounds for CVF 6.6C3 crashes.
I know these compilers are supposed to share code but this was unexpected behaviour.
In case it helps identify the problems, the crashes were associated with "use" statements which use associate other modules which use associated still other modules. The workaround is not to put "use" statements at the module level but in each subroutine individually, and to explicitly specify the variables with "only". Perhaps this is suggestive of a symbol-table overflow.
Keith Refson
Link Copied
1 Reply
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
It's not a help for diagnosis - problems like this require seeing the actual source. There is a lot of shared source between CVF and Intel Visual Fortran, especially in module processing.
6.6C3 is announced on the HP Fortran web site.
Please report any problems you find with Intel Fortran to Intel support.

Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page