Itanium question

Itanium question


I don't know if this is the right place to ask this, but I'm doing an independent study for school on the Itanium and was wondering if anyone is achieving performance increases with Itanium intrinsics or compiler pragmas? What types of apps are you writing and how are they being used?


2 posts / 0 new
Last post
For more complete information about compiler optimizations, see our Optimization Notice.

As your question is broad, I'll be somewhat vague in my attempt to answer.

The intrinsics may see the most use in code which is to be portable between IA32 and IA64, but requires the intrinsics for performance on IA32. There aren't as many situations on IA64 where popular C compilers don't have the ability to engage the performance supported by the architecture (SSE parallel for IA32, SWP for IA64). For applications involving float data, engaging the FTZ intrinsic is particularly important for performance, even if no actual data operations are performed by intrinsics.

Intel C for IA64 has depended strongly on the use of
#pragma ivdep
to engage SWP for loops involving array subscripts. I don't know whether that will continue to be so. Additional optimization pragmas are on the way, so it's clear that some advantage is expected. One such arena is in the optimization of software prefetch instructions.

You might search Intel's IDF site for examples.

Leave a Comment

Please sign in to add a comment. Not a member? Join today