Cyrix Corporation

National Semiconductor Corporation

Douglas Ewing Duschatko

Raul A Garibay, Jr

Jeffrey S Byrne

David A Courtright

Mark W Bluhm

Margaret R Herubin

Marvin Wayne Martinez, Jr

John K Eitrheim

Richard B Reis

Lawrence H Hudepohl

Steve McMahan

Tai Dinh Ngo

A microprocessor is implemented using sense amplifiers to replace CMOS logic circuits, in order to provide low voltage, high frequency switching. The input node of the sense amplifier is maintained at a voltage just above or just below their trip-point of one inverter in order to obtain high-speed switching. Bench mark tests have shown that a microprocessor operating at 2.7 volts may obtain a frequency of 20 MHz and while the same microprocessor may operate at 5.5 volts and 40 MHz.

High speed processor for operation at reduced operating voltage

A write-back coherency system is used, in an exemplary embodiment, to implement write-back caching in an x86 processor installed in a multi-master computer system that does not support a write-back protocol for maintaining coherency between an internal cache and main memory during DMA operations. The write-back coherency system interrupts the normal bus arbitration operation to allow export of dirty data. In response to an arbitration-request (such as HOLD), if the internal cache contains dirty data, the processor is inhibited from providing arbitration-acknowledge (such as HLDA) until the dirty data is exported (the cache is dynamically switched to write-through mode to prevent data in the cache from being made dirty while the bus is arbitrated away). While the requesting bus master is accessing memory, bus snooping is performed and invalidation logic invalidates at least those cache locations corresponding to locations in memory that are affected by the requesting bus master.

Coherency for write-back cache in a system designed for write-through

A write-back coherency system, including FLUSH/INVAL and LOCK protocols, is used, in an exemplary embodiment, in a microprocessor used in a computer system that selectively provides to the processor FLUSH and INVAL signals to implement a limited write-back protocol. The FLUSH/INVAL protocol is used by the computer system to control export and invalidate operations. In response to a FLUSH signal, the microprocessor exports dirty data from the cache. If INVAL is also asserted, the cache is also invalidated (i.e., if FLUSH is asserted and INVAL is not asserted, no invalidation is performed). With the LOCK protocol, LOCKed reads are serviced out of the cache for read hits--however, to maintain compatibility with computer systems that expect a LOCK operation to involve a read followed by a write access to external memory, the microprocessor will still run the external LOCKed read cycle, ignoring the returned data.

Burst ordering logic is used, in an exemplary embodiment, to implement an ascending only burst ordering for cache line fills in 486 computer systems while maintaining compatibility with the conventional 486 burst ordering which uses both ascending and descending burst orders depending upon the position of the requested address (critical Dword) within a cache line (conventional 486 burst ordering is illustrated in Table 1 in the Background). The burst ordering logic (60) implements a 1+4 burst ordering for requested addresses that, for conventional 486 burst ordering, would result in a descending burst order (the exemplary 1+4 burst ordering is illustrated in Table 2 in the Specification). The burst ordering logic includes request modification circuitry (64), address modification circuitry (66), and cacheability modification circuitry (68). If the burst ordering logic detects a cacheable requested address that will cause a cache line fill using an ascending burst order, that requested address is passed through to the bus interface unit for servicing as a normal cache line fill. If, however, the requested address would result in a descending burst order, the cacheability modification circuit modifies cacheability to cause the microprocessor to assert PCD (page cache disable), signaling to the system logic that the requested address is noncacheable--the system logic responds with a single non-burst transfer (1+) of the requested address. The request modification circuitry then signals a burst transfer request for a cache line fill, and the address modification circuitry modifies the requested address to provide a bus cycle address that is within the cache line that contains the requested address but will result in an ascending burst order to transfer that cache line (+4).

Burst transfers using an ascending or descending only burst ordering

Shift-based control logic is used, in an exemplary embodiment, to implement in a microprocessor a circular prefetch queue that stores variable length instructions and transfers four instruction bytes at a time to an instruction decoder. The prefetch queue (10) includes a 16 byte sequential prefetch buffer (12). Access to the buffer is controlled by the shift-based control logic (14) which includes shifter logic that defines a four byte transfer window corresponding to an index byte together with the next three bytes in sequence. For each four-byte transfer operation, the shifter logic enables the four bytes within the transfer window to be read out for transfer to the instruction decoder (20). A transfer operation is initiated by the decoder, which presents the shift-based control logic with a bytes-used indicator, or shift increment. The shift increment denotes the number of bytes used by the previous four byte transfer via a four bit, one-hot selection. In response to receiving the shift increment, the control logic shifts the transfer window an amount specified by the shift increment in preparation for the next four byte transfer.

Control logic for a sequential data buffer using byte read-enable lines

A write-back coherency system is used, in an exemplary embodiment, to implement write-back caching in an x86 processor installed in a multi-master computer system that does not support a write-back protocol for maintaining coherency between an internal cache and main memory during DMA operations. The write-back coherency system interrupts the normal bus arbitration operation to allow export of dirty data, and includes an X%DIRTY latency-control function. In response to an arbitration-request (such as HOLD), if the internal cache contains dirty data, the processor is inhibited from providing arbitration-acknowledge (such as HLDA) until the dirty data is exported (the cache is dynamically switched to write-through mode to prevent data in the cache from being made dirty while the bus is arbitrated away). While the requesting bus master is accessing memory, bus snooping is performed and invalidation logic invalidates at least those cache locations corresponding to locations in memory that are affected by the requesting bus master. The X%DIRTY function provides write-back latency control by dynamically switching the cache from write-back to write-through mode if a cache write would cause the number of cache locations containing dirty data to exceed a predetermined maximum percentage of the total number of cache locations.

Growing community of inventors

Plano, TX, United States of America

Douglas Ewing Duschatko