[LLVMbugs] [Bug 11024] llc is taking 2000x time to compile for ARM target than for x86

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Wed Sep 28 01:43:41 PDT 2011


seb <babslachem at gmail.com> changed:

           What    |Removed                     |Added
             Status|RESOLVED                    |REOPENED
         Resolution|WORKSFORME                  |

--- Comment #10 from seb <babslachem at gmail.com> 2011-09-28 03:43:40 CDT ---
OK I just reopen this topic to add this comment. I've checked with trunc and
now it runs in 10 secs, so however this is a 10x improvement, it is still 100x
slower than for x86 target. Here is an excerpt of use of -time-passes option:

   ---User Time---   --System Time--   --User+System--   ---Wall Time---  ---
Name ---
   9.7726 ( 93.1%)   0.0800 ( 71.4%)   9.8526 ( 92.9%)   9.8801 ( 92.9%)  ARM
Instruction Selection

As you can see, now ARM Instruction selection is taking most of the time.
Hope this can help improve LLVM for ARM.
Feel free to close the issue.
Best Regards

Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

More information about the llvm-bugs mailing list