- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I'm building an app that is run on a cluster of quad CPU nodes. Each CPU is a Xeon 5150 @ 2.66 GHz. If I give the combination of compiler options -O3 -xP -tpp7 -ip, the job dies with a complaint about not being built for the processor. The application is normally built with the same set of options except for -xW in place of -xP.
The bizarre thing to me is that I could build and run the app using -O3 -static -xP -tpp7. Does -xP and -ip in combination cause a problem?
Les
Link Copied
2 Replies
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
-tpp7 would be redundant with -xW or -xP, and may be meaningless with compilers which support those options. Recent compilers have some form of -ip set by default, so you may not require that option. -xO generates code similar to -xP, without the CPU identity check.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Quoting - lhartzman
I'm building an app that is run on a cluster of quad CPU nodes. Each CPU is a Xeon 5150 @ 2.66 GHz. If I give the combination of compiler options -O3 -xP -tpp7 -ip, the job dies with a complaint about not being built for the processor. The application is normally built with the same set of options except for -xW in place of -xP.
The bizarre thing to me is that I could build and run the app using -O3 -static -xP -tpp7. Does -xP and -ip in combination cause a problem?
Les
The difference here is unexpected. Could there be a difference in whether -xP is used to compile the main() routine? I would recommend submitting an issue to Intel Premier Support on this.
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