This instruction computes a 16×16-bit multiply with the lower halfword of the first operand (Imm16 or si4, which can be a register or an immediate value), and the lower halfword of the second operand (snd), giving a 32-bit result (the largest possible size for YASEP32).
See also MUL8H for the 8×8 bits version and MUL16H for the higher halfword.