- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I noticed that latest ICC 10.1 generates long conditional jumps where short ones would suffice even if you write jz short label (I am talking about inline assembler code). For example it spits out 0F 84 20 00 00 00 instead of 74 24 which clearly doesn't make any sense. Any ideas?
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Can someone from Intel please tell me if what I am seeing here is intended behavior or not? Has anyone else seen this?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Nice to see that there is no one around to help me a bit when I need it.
That fact combined with the latest bugs introduced into the forum software by incapable administrators (no edit button, editing kills attachment,
font not fixed width) really kills my motivation to hang around here and help other people.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Igor,
We've tried several cases trying to duplicate the problem, but no success. Could you send yourtestcase? andwhat the compile options used. If you like, you can send to Premier Support
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Igor,
We've tried several cases trying to duplicate the problem, but no success. Could you send yourtestcase? andwhat the compile options used. If you like, you can send to Premier Support for
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Igor,
We've tried several cases trying to duplicate the problem, but no success. Could you send yourtestcase? andwhat the compile options used. If you like, you can send to Premier Support for secority r
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Igor,
We've tried several cases trying to duplicate the problem, but no success. Could you send yourtestcase? andwhat the compile options used. If you like, you can send to Premier Support for security reason.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Here you go, number of NOPs between jumps corresponds with actual instruction sequence lengths in our code. Should work with any compiler options, I have tested using ICC 10.1.021 IA32 for Windows. Use /c /FAc to get the .cod file and check those JE encodings. I have submitted it to the Premier support (#481387).
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page