[llvm-bugs] [Bug 47573] New: [AVX512] Supoptimal mask operation

via llvm-bugs llvm-bugs at lists.llvm.org
Fri Sep 18 00:30:13 PDT 2020


https://bugs.llvm.org/show_bug.cgi?id=47573

            Bug ID: 47573
           Summary: [AVX512] Supoptimal mask operation
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Backend: X86
          Assignee: unassignedbugs at nondot.org
          Reporter: david.bolvansky at gmail.com
                CC: craig.topper at gmail.com, llvm-bugs at lists.llvm.org,
                    llvm-dev at redking.me.uk, spatel+llvm at rotateright.com

-Ofast -mavx512bw -mno-avx512dq -mno-avx512vl -mno-xop
-mprefer-vector-width=512

Clang:
f1(char*, char*, char*):                            # @f1(char*, char*, char*)
        vmovdqu64       zmm0, zmmword ptr [rsi]
        vpcmpltb        k1, zmm0, zmmword ptr [rdx]
        vmovdqu8        zmm1 {k1} {z}, zmmword ptr [rdi]
        vmovdqu8        zmm0 {k1}, zmm1
        vmovdqu64       zmmword ptr [rdi], zmm0
        vzeroupper
        ret
f2(unsigned char*, unsigned char*, unsigned char*):                           
# @f2(unsigned char*, unsigned char*, unsigned char*)
        vmovdqu64       zmm0, zmmword ptr [rsi]
        vpcmpltub       k1, zmm0, zmmword ptr [rdx]
        vmovdqu8        zmm1 {k1} {z}, zmmword ptr [rdi]
        vmovdqu8        zmm0 {k1}, zmm1
        vmovdqu64       zmmword ptr [rdi], zmm0
        vzeroupper
        ret
f3(char*, char*, char*):                            # @f3(char*, char*, char*)
        vmovdqu64       zmm0, zmmword ptr [rsi]
        vpcmpgtb        k1, zmm0, zmmword ptr [rdx]
        vmovdqu8        zmm1 {k1} {z}, zmmword ptr [rdi]
        vmovdqu8        zmm0 {k1}, zmm1
        vmovdqu64       zmmword ptr [rdi], zmm0
        vzeroupper
        ret


GCC:
f1(char*, char*, char*):
        vmovdqu8        zmm0, ZMMWORD PTR [rsi]
        vpcmpb  k1, zmm0, ZMMWORD PTR [rdx], 5
        vmovdqu8        ZMMWORD PTR [rdi]{k1}, zmm0
        vzeroupper
        ret
f2(unsigned char*, unsigned char*, unsigned char*):
        vmovdqu8        zmm0, ZMMWORD PTR [rsi]
        vpcmpub k1, zmm0, ZMMWORD PTR [rdx], 5
        vmovdqu8        ZMMWORD PTR [rdi]{k1}, zmm0
        vzeroupper
        ret
f3(char*, char*, char*):
        vmovdqu8        zmm0, ZMMWORD PTR [rsi]
        vpcmpb  k1, zmm0, ZMMWORD PTR [rdx], 2
        vmovdqu8        ZMMWORD PTR [rdi]{k1}, zmm0
        vzeroupper
        ret

More examples: https://godbolt.org/z/MjaKeE

-- 
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/20200918/91f411e3/attachment.html>


More information about the llvm-bugs mailing list