Intel Corporation

Elliot D Garbus

Byron R Gillespie

Ravi S Rao

William T Futral

Dennis M O'Connor

Jay S Heeb

Barry R Davis

Thomas M Johnson

Mitchell A Kahn

Kenneth C Creta

James W Alexander

Joseph Murray

Marc A Goldschmidt

Nick G Eskandari

Peter Sankhagowit

Douglas D Yoder

Lionel S Smith, Jr

Larry Bates

Dinesh Ranganathan

Real-time performance monitoring facility in an integrated circuit (IC) data processor for monitoring events related to different bus activity. The monitoring facility is accessible via a bus connection the IC. Events include device acquisition and ownership time, and the number of requests and grants on a given bus. The events are counted as occurrences and durations by a number of event counters integrated in the IC. The IC can notify software when the counters overflow. The IC may feature multiple clock domains, including, for instance, multiple bus interfaces operating at different clock frequencies, in which events from different clock domains may be tracked by the same counter. In one embodiment, the performance monitoring facility is integrated into an I/O processor (IOP) die that complies with the popular intelligent I/O (I O) and Peripheral Components Interconnect (PCI) specifications.

Integrated real-time performance monitoring facility

A 32-bit mode operation for a typical 64-bit ECC memory subsystem. In 32-bit mode, each data block will have an 8-bit ECC value, which is consistent with ECC values generated for 64-bit data. This is achieved by prefixing the data with 32 zeroes. When reading out data, memory faults can be corrected and detected using ECC techniques on a zero prefixed data block that is read out of the memory. This allows a memory subsystem to be optimized for bandwidth and latency depending upon this application.

32-bit mode for a 64-bit ECC capable memory subsystem

A message unit that provides a hardware queue interface between a host processor and a local processor handling I/O operations in an I/O platform. Circuitry manages the head and tail pointers of an inbound free queue, an inbound post queue, an outbound free queue and an outbound post queue. Circuitry is also provided for enabling a host processor or bus agent to access these queues in a single bus transaction by reading or writing inbound port registers or outbound port registers. The queue elements contain handles of message buffers. The invention automatically performs the specific task of locating the next element in a queue, altering that element, and modifying a queue descriptor (i.e., a head or a tail pointer) to indicate the next element for a next queue access. A plurality of registers are used for selectively interrupting either the host processor or the local processor when the queues are written to by either the host processor, a bus agent, or the local processor.

Method and apparatus for transporting messages between processors in a

A method and apparatus for monitoring an event is disclosed. In one embodiment, a ratio counter is provided which includes a first counter having a first count and a second counter having a second count. The method increments the first count of the first counter on an occurrence of an event by a first device and shifts the ratio counter to the right by a predetermined number of bits when the first count reaches a maximum count. The method further takes a ratio between the first count and the second count where the ratio indicates the relative occurrence of the event by the first device and a second device. In another embodiment, the method decrements the second count of the second counter when the first count reaches a maximum count. In yet another embodiment, the method decrements the first count of the first counter and the second count of the second counter when the first count reaches a maximum count.

Method and apparatus for providing scaled ratio counters to obtain agent

An application accelerator unit (AAU) that is integrated as part of a data processor, such as an I/O processor (IOP) integrated circuit. In one embodiment, the AAU includes logic for improving the performance of storage applications such as Redundant Array of Inexpensive Disks (RAID). The AAU performs boolean operations such as exclusive-or (XOR) on multiple blocks of data to form the image parity block which is then written to the redundant disk array. Additionally, the AAU may feature adder logic configured to perform an addition such as a network header checksum calculation on each data packet. The AAU includes a memory-mapped programming interface that allows software executed by a core processor in the IOP to utilize the AAU for accelerating storage and networking applications as well as for local memory DMA-type transfers, using the chain descriptor construct.

Data processor having integrated boolean and adder logic for

A multi-functional device that integrates a high performance processor into a PCI to PCI bus bridge. The invention consolidates a high performance processor, a PCI to PCI bus bridge, PCI bus-processor address translation unit, direct memory acces's (DMA) controller, memory controller, secondary PCI bus arbitration unit, inter-integrated circuit (I.sup.2 C) bus interface unit, advanced programmable interrupt (APIC) bus interface unit, and a messaging unit into a single system which utilizes a local memory. The PCI bus is an industry standard high performance, low latency system bus. The PCI to PCI bridge provides a connection path between two independent 32-bit PCI buses and provides the ability to overcome PCI electrical loading limits. The addition of the local processor brings intelligence to the PCI bus bridge.

Architecture for an I/O processor that integrates a PCI to PCI bridge

A wait state mechanism for lengthening a microprocessor's bus cycle to allow data transfers between slower off-chip devices. A microprocessor is responsive to a bus control signal generated by external programmable logic which instructs the microprocessor to insert wait states of varying number depending on the component involved in a bus transaction. The microprocessor receives only a single input from the programmable logic and varies its bus cycle length accordingly.

Multi-function microprocessor wait state mechanism using external

Dynamic appending of chain descriptors is described with reference to a computer system having a host processor, a DMA unit, a host memory and an external memory wherein the DMA unit controls transference of data between the host memory and the external memory based upon data transference parameters specified in chain descriptors created by the host processor and stored as data structures within the host memory. In accordance with one method and apparatus described herein, dynamic appending of chain descriptors is achieved by employing a resume bit stored within a register of the DMA unit. The host processor, upon creating a new group of chain descriptors to be appended to a previous group, updates a link value within a last chain descriptor of the previous group to point to the first chain descriptor of the new group and also sets the resume bit within the DMA unit. The DMA unit reads chain descriptor parameters, including link values, they perform a data transfer operation specified by the chain descriptor parameters. Upon completion of the transfer operation, the DMA unit examines the resume bit and, if set, the DMA unit rereads the link value for the current chain descriptor. If the resume bit has not been reset, the DMA unit merely proceeds to process the next chain descriptor specified by the previously read link value or, if that link value is a null value, the DMA unit merely terminates operation. In an alternative embodiment described herein, the host processor updates link values but does not set a resume bit within the DMA unit. Rather, the DMA unit initially reads all parameters for a chain descriptor except for the link value. Then, after completion of the data transfer operation specified by the chain descriptor, the data unit reaccesses the chain descriptor to read the link value. Hence, if the link value is updated by the host processor while the DMA unit is processing a chain descriptor, the DMA unit will nevertheless access the updated link value upon completion of the data transference operation. Method and apparatus embodiments are described herein.

System for creating new group of chain descriptors by updating link

The system and method described provide for the detection and protection of memory accesses without the overhead typically incurred by memory management units. The processor includes a guarded memory unit, which monitors memory accesses to be performed by monitoring transmissions across the memory bus. The guarded memory unit includes a plurality of registers which identify memory addresses and modes which can cause a memory protection or detection violation to occur. If a memory protection violation occurs, a cancel signal is issued to cancel the memory operation prior to completion in order to protect the memory from unauthorized accesses. If a memory violation is detected, the memory operation is permitted to complete and a fault signal is issued to the processor to identify that a memory violation has been detected. As the structure of the protection mechanism does not require separate cycles in the processor, and simply monitors the memory bus for memory accesses, memory protection and detection can be performed with no additional overhead at the processor.

System for protecting memory accesses by comparing the upper and lower

System for protecting unauthorized memory accesses by comparing base

A micro processor emulator in which a set of core micro processor registers are the communication interface between an external system and a core-ported memory. The registers are connected to a serial scan port for transfer of information between a halted emulation environment and the external system. The serial port includes a command register that receives a jump address to initiate execution of a software monitor. Two special bits are provided in the command register, one that indicates a break, and one that indicates a Fast Break GO. This provides a break mechanism for a micro processor chip which does not have a dedicated memory bus. This break mechanism is the mechanism by which a halt or an asynchronous break is effected. After a fast break, the Fast Break GO mechanism does the action described by one command, and then immediately goes back to emulation without any external processor intervention.

Growing community of inventors

Scotsdale, AZ, United States of America

Elliot D Garbus