摘要:
A method and system for processing instructions in a floating point unit for executing denormalized numbers in floating point pipeline via serializing uses an instruction unit and having a control unit and a pipelined data flow unit, a shifter and a rounding unit. The floating point unit has an external feedback path for providing intermediate result data from said rounding unit to an input of the pipelined data flow unit to reuse the pipeline for denormalization by passing intermediate results in the pipeline which have a denormalized condition computed after the exponent calculation of the shifting circuit directly from the rounding unit to the top of the dataflow in the pipeline via an external feedback path. The pipelined has two paths which are selected based on the presence of other instructions in the pipeline. If no other instructions are in the pipeline a first path is taken which uses the external feedback path from the rounding unit back into the top of the dataflow. When there are instructions in the pipeline a shifter unit performing normalization of the fraction indicates possible underflow of the exponent, and prepares to hold the exponent and the fraction in a floating point data flow register; and upon detection of exponent underflow during the rounder stage and detection of any other instructions in pipeline; then the control unit forces an interrupt for serialization, and cancels execution of this instruction and other instructions in pipeline.
摘要:
The present invention is a variable-delay division (VDD) scheme implementable in hardware to execute signed and unsigned integer division and remainder operations in digital processor. The VDD scheme advantageously uses hardware utilized for multiplication to implement a 2-bits/cycle alignment step to iteratively align the divisor with the dividend. This speeds up the alignment phase of integer division. Quotient bits are produced at the rate of 1-bit/cycle using the well-known restoring scheme. For 32-bit 2's complement operands, the scheme has a delay less than a fixed-delay scheme for most operands.
摘要:
A combined multiplier/shifter (150) uses an existing high-speed multiplier (55) to perform both multiplies and programmable left and right shifts without a dedicated high-speed shifter. A shift decoder (160) used in a shift mode provides first recoded signals according to a shift count and a shift direction. A recoder (161) recodes a multiplier input in a multiply mode to provide second recoded signals. A multiplier array (163) receives either a multiplicand or a shift operand at its multiplicand input, and uses either the first or second recoded signals selectively according to the mode. An output of the multiplier array (163) is either a product in the multiply mode or a first shift result in the shift mode. An output shifter (157) selectively adjusts the first shift result according to the shift direction to provide a second, final shift result.
摘要:
A circuit[100] for shifting the bits of an X word to obtain a Y word which is rounded to the nearest odd integer if any bit having the value 1 was shifted off of the word during the shifting operation. The circuit avoids biasing in the integer rounding operation. The shifting operations are accomplished with the aid of multiplexing circuits[101-105]. The rounding operation is accomplished with the aid of multiplexing circuits[105] that connect the least significant bit of Y to (X₀ OR X₁ OR ...X m ), where m is the number of places by which X is shifted.
摘要翻译:一个用于移位X字的位的电路[100],以获得一个Y字,该Y字被舍入到最近的奇整数,如果任何具有值1的位在移位操作期间被偏离了该字。 电路避免整数舍入操作中的偏置。 借助于复用电路[101-105]来实现移位操作。 借助于将Y的最低有效位连接到(X0 OR X1 OR ... Xm)的多路复用电路[105]来完成舍入操作,其中m是X移位的位置数。
摘要:
A data shifting circuit comprises a barrel shifter (11) for shifting by a plurality of bits data having a width twice that of a certain data width, and a data controller for supplying the same data having the certain data width commonly to the most significant bits and the least significant bits of the barrel shifter means (11).
摘要:
A high-speed barrel shifter (20) includes a shifter array (25) having a matrix of transistors (40) located at intersections of rows and columns of the matrix (40). The rows and columns alternately function as source and destination terminals. A data-dependent fill portion (48) fills a data-dependent value such as a sign bit into vacated bit positions along rows in a bottom left portion (42). Thus the barrel shifter (20) performs a data-dependent fill instruction within the shifter array (25) and avoids extra clock cycles associated with post-array processing. In one embodiment, an isolation portion (44, 45) separates a top right portion (41) of the matrix (40) from the bottom left portion (42) along a diagonal (43). The isolation portion (44, 45) isolates transistors in the bottom left portion (42), which are associated with rotates and fills, from transistors in the top right portion (41), which are associated with shifts, according to the direction of the shift.