Abstract:
PROBLEM TO BE SOLVED: To reduce the number of computing elements constituting a square multiplier for floating-point numbers and also to increase the processing speed by circuit techniques for an arithmetic unit. SOLUTION: The square multiplier for floating-point numbers is provided with a false carry generation circuit 21a which falsely generates information related to carry of prescribed bits in the arithmetic of a variable as an arithmetic object, an MSB(Most Significant Bit) look-ahead circuit 21b which decides the MSB in the arithmetic result in advance from the variable as the arithmetic object, and a combinational circuit which uses information related to carry, which is generated by the false carry generation circuit 21a, to perform arithmetic of the variable while performing rounding processing based on the position of the MSB decided by the MSB look-ahead circuit 21b.
Abstract:
PROBLEM TO BE SOLVED: To enable a receiving scan latch to release original data, even if the arrival times for the release/capture clocks differ in each clock domain, in data transfer between clock domains. SOLUTION: A plurality of clock domains to each of which a testing clock is supplied from a separate clock generation circuit are provided. In each clock domain, scan latch on the border of the clock domains for receiving an input from another clock domain includes a master latch 30 for latching the input in response to a first clock; a slave latch 32 for latching an output from the master latch in response to a second clock; a selector 34 for supplying a scan input to the master latch, when a mode switchover signal is at a first level and supplying a system input to the master latch, when the signal is at a second level; and a clock control circuit 36 for turning off the first clock, when the mode switchover signal transits from the first level to the second level. COPYRIGHT: (C)2007,JPO&INPIT
Abstract:
PROBLEM TO BE SOLVED: To provide an information processor, for which it is not necessary to analyze the number of stages of FIFO from the character of data for improving performance, by dynamically having the number of stages of FIFO optimal at that time. SOLUTION: This device is provided with a data FIFO 22 for storing data sets and a next pointer 29 having storage areas as many as in this data FIFO. The preceding data set is stored in a storage area '1' of the data FIFO 22 and the following data set is stored in a storage area '7' of the data FIFO 22. At this time, '7' is stored in the storage area '1' of the next pointer 29 as storage area information of following data. On the basis of this storage area information '7', the following data set is read out of the storage area '7' of the data FIFO 22.