- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Is there a vector single-float complex multiply routine in MKL? I looked in the BLAS section and VML and elsewhere without finding it (there is a divide, but no mult!) Am I just missing something simple? My naive C++ method is too slow, and icc doesn't vectorize it.
Link Copied
1 Reply
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Dear Gary,
I would be interested to see your naive C++ methodto investigate why icc does not vectorize it. I suspect you separated the real and imaginary parts into explicit operations of floats, and the Intel compiler is not able to map these constructs back into an intermediate on complex operations. The vectorizer performs much better on the complex data types of C99, but this is restricted to C source for now. Perhaps you will find this thread helpful:
http://software.intel.com/en-us/forums//topic/46952
Hope this helps,
Aart Bik
http://www.aartbik.com/
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