[LLVMbugs] [Bug 10496] [AVX] missing intrinsics for integer min/max instructions

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=10496

Bruno Cardoso Lopes <bruno.cardoso at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID

--- Comment #3 from Bruno Cardoso Lopes <bruno.cardoso at gmail.com> 2011-08-02 11:18:43 CDT ---
We don't want to duplicate intrinsics, so we just match x86_sse2* to the right
sse/avx instruction later. Both VPMINUB and PMINUB for example, are matched
from int_x86_sse2_pminu_b. If you check other compilers you will see that they
don't define a new builtin for every new AVX encoded 128-bit instruction, we do
the same with the builtins and the IR intrinsics. You can use
int_x86_sse2_pminu_b in your generated IR, and when specified -mattr=+avx, you
get the right instruction. If it happens that some of them won't select the
right instruction, then let me know cause it's a bug! We have tests to check
all of them though, look at test/CodeGen/X86/avx-intrinsics-x86.ll

-- 
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.