4 ms·

Yeah. The problem with them is that using double-double based methods gives you 104 bits of precision that works with vectorization (and therefore is faster tha

Yeah. The problem with them is that using double-double based methods gives you 104 bits of precision that works with vectorization (and therefore is faster than x87).

gpderetta 5y ago
I assume double-double would be the emulated __float128 type in GCC (or hand written equivalent)? Can GCC vectorize it or it needs to be done by hand?
- jcranmer 5y ago
  __float128 is emulated quad-precision, I think.
  - gpderetta 5y ago
    Not exactly, the mantissa is increased, but the exponent is the same as standard double precision, so it is not an IEEE quad.