[llvm-bugs] [Bug 39137] Imrove perfromance when user sets same number of threads

via llvm-bugs llvm-bugs at lists.llvm.org
Wed Nov 14 06:07:06 PST 2018


Andrey Churbanov <andrey.churbanov at intel.com> changed:

           What    |Removed                     |Added
 Fixed By Commit(s)|                            |r346862
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #1 from Andrey Churbanov <andrey.churbanov at intel.com> ---
The function omp_set_num_threads() fixed to not write into internal structure
if requested number of threads is the same as the value of internal control
variable (ICV).

With num_threads() clause things are more complicated, because the write here
is cheaper and this write eliminates the need to read ICV twice - here and in
the following __kmp_fork_call. Attempt to avoid write op here caused some
performance regressions in our testing, so I'd prefer to leave the clause
implementation intact.

You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20181114/d921dd5a/attachment.html>

More information about the llvm-bugs mailing list