[llvm-bugs] [Bug 45003] std::complex support in offload region needed

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=45003

Ye Luo <xw111luoye at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|CONFIRMED                   |RESOLVED
         Resolution|---                         |FIXED

--- Comment #8 from Ye Luo <xw111luoye at gmail.com> ---
(In reply to Ye Luo from comment #7)
> As of today.
> 
> PASS: $ clang++ -fopenmp -fopenmp-targets=nvptx64 complex.cpp && ./a.out
> PASS: $ clang++ -fopenmp -fopenmp-targets=nvptx64 -O3 complex.cpp && ./a.out
> PASS: $ clang++ -fopenmp -fopenmp-targets=nvptx64 -O3 -ffast-math
> complex.cpp && ./a.out
> 
> FAIL: $ clang++ -g -fopenmp -fopenmp-targets=nvptx64 complex.cpp && ./a.out
> PASS: $ clang++ -g -fopenmp -fopenmp-targets=nvptx64 -O3 complex.cpp &&
> ./a.out
> PASS: $ clang++ -g -fopenmp -fopenmp-targets=nvptx64 -O3 -ffast-math
> complex.cpp && ./a.out
> 
> cuda-memcheck indicates invalid memory access in the failed run.
> It happens when T is float and the division involves both complex and real
> type.
>   test_div<T>(std::complex<T>(0, 1), T(0.5));
>   test_div<T>(T(0.5), std::complex<T>(0, 1));

I cannot reproduce this issue any more with both 11RC1 and today's master
branch.

-- 
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/20200816/1c30962a/attachment.html>