Posits: new approach could be much more efficient than IEEE 754 floating-point computation Article 1380 words 7 votes