triton.language.clmul

triton.language.clmul(x, y, _semantic=None)

Computes the element-wise carry-less product of x and y.

The operands are interpreted as binary polynomials over GF(2), so additions in the multiplication are replaced by XORs. The result contains the low N bits of the product, where N is the bit width after integer type promotion. Signed operands are interpreted by their bit patterns. The inputs must be integers and follow the usual broadcasting and type promotion rules.

Parameters:
  • x – the first input

  • y – the second input