Am I understanding correctly that Open64 can not compile code if AVX intrinsics are used in source file but it can use AVX intrnsics if it auto-vectorizes?
Yes you are right. These intrinsics feature support is currently under developement and will be available in the up coming release soon.
We have new version of the Open64 compiler, v4.5.2, released with performance improvements and many bug fixes. The issue you reported has also been fixed. Please check.
For defails refer http://developer.amd.com/tools/open64/pages/default.aspx
Thank you santosh. Does AMD have mailing lists for sending update info for their SDKs etc. ? That would be awesome because I do make packages for computing clusters and well it is easier for me to track down what is updated....
The best resource for news about updates to SDK and other AMD tools is the hcNewsFlash email newsletter. You can see old versions & sign up to get the newsletter here: http://developer.amd.com/community/newsletter/Pages/default.aspx
Retrieving data ...