clang++: abs of float vector comes out zero

[kfjahnke]

clang++ (clang version 10.0.0-4ubuntu1) yields zero when abs() is called with a float vector. Here's code to reproduce the error:

#include <iostream>
#include <experimental/simd>

int main ( int argc , char * argv[] )
{
  std::experimental::simd
          < float ,
            std::experimental::simd_abi::fixed_size < 2 >
          > x ;

  x[0] = -1.0f ;
  x[1] =  1.0f ;

  auto y = abs ( x ) ;

  for ( int i = 0 ; i < 2 ; i++ )
  {
    std::cout << "x[" << i << "] : " << x[i] << " -> " ;
    std::cout << "y[" << i << "] : " << y[i] << std::endl ;
  }
}

Actual Results

compiling with clang I get the error:

$ clang++ -std=c++17 stdsimd_abs.cc
$ ./a.out
x[0] : -1 -> y[0] : 0
x[1] : 1 -> y[1] : 0

Expected Results

g++ yields the correct result.

$ g++ -std=c++17 stdsimd_abs.cc
$ ./a.out
x[0] : -1 -> y[0] : 1
x[1] : 1 -> y[1] : 1

Kay