For those who found parts of our description of the forthcoming Intel Corp N-10 microprocessor hard going (CI No 1,119), all we can say is you didn’t see the material from which we were working – try this passage from Intel’s description of the way the N-10’s floating point multiplier works: Since the upper-product bits are not yet available, it predicts if the product is in [1,2] or [2,4] and the associated round and sticky bits; a freeze is invoked if the product range cannot be predicted without a full carry propagate add of the upper product bits; the third stage does a carry propagate add of the upper product bits and the round bit.
Content from our partners
