Digital Equipment Corporation

Digital Euipment Corporation

William C Madden

Vidya Rajagopalan

William Ralph Wheeler

Sridhar Samudrala

Douglas E Sanders

G Michael Uhler

George M Uhler

William J Grundmann

William J Bowhill

A floating point multiply of two n-bit operands creams a 2n-bit result, but ordinarily only n-bit precision is needed, so rounding is performed. Some rounding algorithms require the knowledge of the presence of any '1' in the n-2 low-order bits of the 2n-bit result. The presence of such a '1', indicates the so-called 'sticky bit' is set. The sticky bit is calculated in a path separate from the multiply operation, so the n-2 least significant sums need not be calculated. This saves time and circuitry in an array multiplier, for example. In an example method, the difference between n and the number of trailing zeros, 'x', in one of the n-bit operands is detected, by transposing the operand and detecting the leading one. The other operand is right-shifted by a number of bit positions equal to this difference. A sticky bit is generated if any logic '1's' are in the low-order n-x-2 bits fight shifted out of the second operand.

Method and apparatus for controlling a rounding operation in a floating

A master/slave latch circuit employs a single-wire clock, with the clock being applied to only N-channel transistors in the master latch and to only P-channel transistors in the slave latch so that a race-through condition is alleviated in the event of clock skew. The circuits are of ratioless operation, since P-channel transistors are used in each circuit to pull the high side to the supply voltage, and N-channel transistors are used on the low side to assure a zero voltage level. Input to each latch is to the gates of a P-channel pull-up and an N-channel pull-down, while the storage node is between the two clocked transistors. The level of the storage node is inverted and fed back to at transistor across one of the clocked transistors, the one on the high side for the master latch and the low side for the slave latch, and these feedback transistors are of a channel type to support the ratioless scheme.

Two-stage CMOS latch with single-wire clock

To reduce the processing time required for correcting a fault, the instruction decorder segment and the first execution segment of a pipelined processor are provided with 'state silos' that are operative during normal instruction execution to save a sufficient amount of state information to immediately restart the instruction decoder segment and the first execution segment by reloading the state information having been stored in the state silos. The state silos, for example, include a queue of registers clocked by a common clocking signal that is inhibited during correction of the fault. When the fault is corrected, multiplexers select the state information from the silos to be used by the respective pipeline segments. In a preferred embodiment, the instruction decoder segment decodes variable length macroinstructions into operand specifiers and operations to perform upon the specifiers. The first execution segment receives control information when a new operand specifier or operation is decoded, and otherwise holds the previously received control information. A microsequencer issues a series of microinstructions for each specifier or operation having been decoded, and also issues a series of microinstructions in a fault routine when a fault occurs. The microsequencer is also provided with a state silo so that the normal sequence of microinstruction execution is resumed when the fault is corrected.

Application of state silos for recovery from memory management exceptions

A memory stack used for storing microinstruction addresses in a pipelined CPU is constructed as a last-in, first-out memory using a stack pointer which applies a read control to one location of the stack and applies a write control to the next higher location. An unconditional read and write is done every machine cycle, before a microinstruction could be decoded, then the data on the read bus, or data from the write bus, is used and the pointer is incremented or decremented if a stack Push or Pop is decoded. These correspond to a Call or Return microinstruction. Thus the delay in decoding the microinstruction does not prevent completion of the stack operation in one machine cycle.

Microinstruction addressing in high-speed CPU

A general purpose sense amplifier, suited for memory and level shifting applications, is provided. The present invention provides a high input impedence for less loading of bit line voltages, wherein operation is relatively insensitive to capacitive mismatches on input bit line pairs. Inherent in the high input impedence design is the built-in isolation between input and output circuitry. The present invention also provides a full rail to rail separation of the output bit line voltages without requiring additional pull-up or pull-down circuitry. The present invention also provides a single strobing input for activating and deactivating the sense amplifier. The present invention also provides minimal circuitry with high speed characteristics and low power dissipation.

Growing community of inventors

Lexington, MA, United States of America

William C Madden