<html>
    <head>
      <base href="https://bugs.llvm.org/">
    </head>
    <body><table border="1" cellspacing="0" cellpadding="8">
        <tr>
          <th>Bug ID</th>
          <td><a class="bz_bug_link 
          bz_status_NEW "
   title="NEW - Suboptimal SLP vectorization"
   href="https://bugs.llvm.org/show_bug.cgi?id=50555">50555</a>
          </td>
        </tr>

        <tr>
          <th>Summary</th>
          <td>Suboptimal SLP vectorization
          </td>
        </tr>

        <tr>
          <th>Product</th>
          <td>libraries
          </td>
        </tr>

        <tr>
          <th>Version</th>
          <td>trunk
          </td>
        </tr>

        <tr>
          <th>Hardware</th>
          <td>PC
          </td>
        </tr>

        <tr>
          <th>OS</th>
          <td>Linux
          </td>
        </tr>

        <tr>
          <th>Status</th>
          <td>NEW
          </td>
        </tr>

        <tr>
          <th>Severity</th>
          <td>enhancement
          </td>
        </tr>

        <tr>
          <th>Priority</th>
          <td>P
          </td>
        </tr>

        <tr>
          <th>Component</th>
          <td>Scalar Optimizations
          </td>
        </tr>

        <tr>
          <th>Assignee</th>
          <td>unassignedbugs@nondot.org
          </td>
        </tr>

        <tr>
          <th>Reporter</th>
          <td>david.bolvansky@gmail.com
          </td>
        </tr>

        <tr>
          <th>CC</th>
          <td>llvm-bugs@lists.llvm.org
          </td>
        </tr></table>
      <p>
        <div>
        <pre>#define N 16

void f (unsigned short *restrict a, unsigned char *restrict b,
   unsigned char *restrict c)
{
  for (int i = 0; i < N; ++i)
    {
      unsigned short res = b[i] + c[i];
      a[i] = (res + (res >> 1)) >> 2;
    }
}



<a href="https://godbolt.org/z/1x8hPbo8d">https://godbolt.org/z/1x8hPbo8d</a>

LLVM -O3 -mavx:
f:                                      # @f
        vpmovzxbd       xmm0, dword ptr [rsi + 4] # xmm0 =
mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
        vpmovzxbd       xmm1, dword ptr [rsi]   # xmm1 =
mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
        vpmovzxbd       xmm2, dword ptr [rdx + 4] # xmm2 =
mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
        vpmovzxbd       xmm3, dword ptr [rdx]   # xmm3 =
mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
        vpaddd  xmm0, xmm2, xmm0
        vpaddd  xmm1, xmm3, xmm1
        vpsrld  xmm2, xmm0, 1
        vpsrld  xmm3, xmm1, 1
        vpaddd  xmm1, xmm3, xmm1
        vpaddd  xmm0, xmm2, xmm0
        vpsrld  xmm0, xmm0, 2
        vpsrld  xmm1, xmm1, 2
        vpackusdw       xmm0, xmm1, xmm0
        vmovdqu xmmword ptr [rdi], xmm0
        vpmovzxbd       xmm0, dword ptr [rsi + 12] # xmm0 =
mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
        vpmovzxbd       xmm1, dword ptr [rsi + 8] # xmm1 =
mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
        vpmovzxbd       xmm2, dword ptr [rdx + 12] # xmm2 =
mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
        vpaddd  xmm0, xmm2, xmm0
        vpmovzxbd       xmm2, dword ptr [rdx + 8] # xmm2 =
mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
        vpaddd  xmm1, xmm2, xmm1
        vpsrld  xmm2, xmm0, 1
        vpsrld  xmm3, xmm1, 1
        vpaddd  xmm1, xmm3, xmm1
        vpaddd  xmm0, xmm2, xmm0
        vpsrld  xmm0, xmm0, 2
        vpsrld  xmm1, xmm1, 2
        vpackusdw       xmm0, xmm1, xmm0
        vmovdqu xmmword ptr [rdi + 16], xmm0
        ret


GCC -O3 -mavx:
f:
        vmovdqu xmm1, XMMWORD PTR [rsi]
        vmovdqu xmm0, XMMWORD PTR [rdx]
        vpmovzxbw       xmm3, xmm1
        vpsrldq xmm1, xmm1, 8
        vpmovzxbw       xmm2, xmm0
        vpsrldq xmm0, xmm0, 8
        vpaddw  xmm2, xmm2, xmm3
        vpmovzxbw       xmm1, xmm1
        vpmovzxbw       xmm0, xmm0
        vpaddw  xmm0, xmm0, xmm1
        vpsrlw  xmm1, xmm2, 1
        vpaddw  xmm1, xmm1, xmm2
        vpsrlw  xmm1, xmm1, 2
        vmovdqu XMMWORD PTR [rdi], xmm1
        vpsrlw  xmm1, xmm0, 1
        vpaddw  xmm0, xmm1, xmm0
        vpsrlw  xmm0, xmm0, 2
        vmovdqu XMMWORD PTR [rdi+16], xmm0
        ret


Plain -O3 codegen is suboptimal too. AVX512 codegen looks good.</pre>
        </div>
      </p>


      <hr>
      <span>You are receiving this mail because:</span>

      <ul>
          <li>You are on the CC list for the bug.</li>
      </ul>
    </body>
</html>